A Series of 62 Skull Base Chordomas in Pediatric and Adolescent Patients: Clinical Characteristics, Treatments, and Outcomes
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.294548
Source of Support: None, Conflict of Interest: None
Keywords: Adolescence, chordoma, outcome, pediatric, skull base
Chordomas are rare malignant bone tumors., The age-adjusted incidence rate is 0.08 per 100,000 population with a male preponderance (male: female ratio of 1.7:1). Because of the paucity of skull base chordoma and the complex anatomy of the skull base, the surgical outcome has long been discouraging. Less than 5% of chordomas present in the first two decades. The majority of primary pediatric and adolescent chordomas (PAC) are located in the cranium.,,,, According to the previous literature, the long-term prognosis of these pediatric patients has been inconsistent., We present our center's experience of a series of PAC patients, focusing on the clinical characteristics, surgery, and long-term outcomes. To our knowledge, this report provides data on the largest number of Asian PAC patients to date.
A total of 516 skull base chordoma patients underwent 655 operations between June 2007 and March 2016 in our center. Patients presenting with chordomas in their first two decades were included in this study. Clinical records, radiographic images, and surgery videos of the patients were retrospectively evaluated to confirm the diagnosis, treatment-related parameters, and predictors of outcome. Approval for this study was received from the hospital ethics board on August 14, 2018. Individual patient consent was not sought as this was a descriptive retrospective study with all data anonymized.
All patients underwent MRI and CT scans preoperatively. According to the endoscopic endonasal approach (EEA),,, the clivus was divided into three sections in the sagittal plane using the boundaries of the clival anatomical markers (sellar floor plane and sphenoid sinus floor plane); these sections were the superior, middle, and inferior clivus. Based on the bilateral boundaries, the lines connecting the bilateral medial orbital wall, lateral wall of the cavernous sinus, internal auditory canal, jugular tuberculum, hypoglossal nerve hole, and occipital condyle were used to divide the skull base into the midline and paramedian regions.
Tumor volume was calculated by the formula: Volume = length × width × height/2. Length, width, and height were recorded at the longest point for each of the three dimensions.
Operative notes were reviewed to collect surgical information, including surgical approach, tumor texture, and resection extent. The degree of tumor resection was evaluated by postoperative images and operative notes. GTR was defined as the absence of residual tumor (including the eroded bone). Near-total resection (NTR) was defined as an absence of tumor as observed by microscope or endoscope, with radiological evidence that >90% of the tumor was removed. Subtotal resection (STR) was defined as >70% of the tumor resected, and partial resection (PR) was defined as ≤70% of the tumor resected.
All patients were evaluated three months after surgery for an initial follow-up. After that point, MRI and CT were scheduled at 6-month intervals. The final follow-up for this study was in January 2019 through telephone interviews.
Student's t-test or the Mann-Whitney U test was used to compare two variables. Clinical experience and statistical analysis were used to decide whether continuous variables should be categorized. The χ2 or Fisher exact test was used to compare two proportions; for comparisons between groups of more than two unpaired values, we employed the Kruskal–Wallis H-test. Survival analysis was performed with the Kaplan-Meier method. Progression-free survival (PFS) was calculated from the date of the first surgery, and overall survival (OS) was defined as the time from diagnosis to disease-specific death or the end of observation. Censoring was applied to the last follow-up date of no death occurred or to the death date if death was not disease-specific. Statistical analysis was performed with SPSS (IBM Inc., Armonk, New York). P < 0.05 was considered statistically significant.
A total of 62 PAC patients underwent 79 neurosurgery operations during the 9-year period, which accounts for 12% of all chordoma patients (62/516). There was a female predominance, with a male-to-female ratio of 1:1.38. Among them, 55 patients were diagnosed with primary chordomas, one patient received radiosurgery one year before neurosurgery, and the other six patients had previously undergone surgeries in other facilities. The median age of all the patients when the first operations were performed was 14 years old (range 3–20 years).
The median duration between initial symptoms and diagnosis was 6 months (range 0.5–96 months). Skull base chordomas present with a wide clinical spectrum, including diplopia, headache, vision deterioration, stuffy nose, hemiplegia, ataxia, and cranial nerve palsy. The most common symptom is diplopia (24/79, 30%), followed by headache (19/79, 24.1%) [Table 1]. The patients may present with neck pain and limited movement when the tumor is located in the craniocervical junction area. For tumors protruding into the oropharynx, patients generally exhibit symptoms of snoring.
The image signal presentations were generally typical and easy to diagnose. Of note, one patient's image was similar to that of epidermoid cysts, even diffusion-weighted MRI was not able to diagnose correctly.
Judging from the sagittal view of the images, two cases belonged to the superior clivus type, which invades the cavernous sinus. Thirteen cases belonged to the superior-middle clivus type, seven cases to the middle-inferior clivus type, 24 cases to the whole clivus type, and 33 cases were localized to the craniocervical junction area [Figure 1]a. Judging from the axial view, there were 29 tumors in the midline region and 49 tumors in the midline region with extension laterally, while only one tumor was purely located in the paramedian region [Figure 1]b.
Thirty-three tumors were located extradurally, and the other 46 tumors invaded the clival dura mater and extended intracranially. No primary intradural chordoma was found.
Fifty-eight cases (58/62, 93.5%) were conventional chordomas. Chondroid chordomas comprised 4.8% (3/62) of the cases. One case was dedifferentiated chordoma (1/62, 1.6%).
The extension patterns of tumors were the main factor for choosing the optimal approach, and the surgeons' level of experience was also a key factor for determining approaches. If the tumor was located predominantly in the midline region and the main part invading the intracranial space did not cross the bilateral boundaries of the midline region, EEA, or an endoscopic transoral approach was more often chosen as the first-line approach [Figure 2]. If the tumor invading the intracranial space was large and with extensive lateral extension, open approaches were chosen as the main approach with the aim of adequate decompression of the brain stem and cranial nerves. For open approaches, the principles were as follows: (1) for tumors located in the superior and middle clivus, a presigmoid approach or middle fossa approaches were applied [Figure 3] and (2) the far lateral approach provides access to the middle-inferior clivus, the pontomedullary junction, and the anterolateral foramen magnum [Figure 4].
In summary, 56 open approaches, 17 EEAs, five transoral approaches (four cases with endoscopy; one case with microscopy) and a combined endoscopic endonasal and transoral approach were performed [Table 2]. Fifty patients underwent a single surgery, and the other twelve patients underwent more than one surgery.
Extent of resection
The total resection rate was 20.3% (16/79), NTR 39.2% (31/79), STR 36.7% (29/79), and PR 3.8% (3/79). There were no significant differences in the GTR rate according to approaches (open approach vs. EEA, P = 0.535) and tumor texture (P = 0.346). However, the GTR rate was highest (29%, 9/31) in the soft tumor group and lowest (0%, 0/9) in the hard texture group.
When the tumors were divided into primary and recurrent groups, the GTR rate was not significantly different (P = 0.127) between groups. However, the GTR rate was 25.5% (14/55) in the primary group but only 8.3% (2/24) in the recurrent group.
When tumor location was considered, the GTR rate for tumors located in the superior clivus was highest (50%, 1/2) but was lowest for tumors located in the craniocervical junction area (9.1%, 3/33) [Table 3]. By comparing the GTR rates for tumors located in the craniocervical junction area with those located in other areas, a statistically significant difference was found (P = 0.048).
The median tumor volume was 36,166.5 (range: 1,681.5–168,000) mm3. The tumors were divided into three groups by tumor volume, which included groups with tumor volumes less than 20 000 mm3, between 20,000 and 80,000 mm3 and larger than 80,000 mm3. The tumor volumes were significantly different (P < 0.001) among different tumor locations. The percentage of large tumors (larger than 80,000 mm3) was highest (27.3%, 9/33) in the craniocervical junction area. The GTR rates were different among the three tumor volume groups (P = 0.034): 34.5% (10/29) in the less than 20,000 mm3 group, 15.0% (6/40) in the 20,000–80,000 mm3 group and 0% (0/10) in the larger than 80,000 mm3 group.
Residual tumors were found in 19 cases in the retropharyngeal space, 14 cases in the cavernous sinus, 13 cases in extracranial parts of the jugular foramen, eight cases in the petrous apex and Meckel's cave, seven cases in the posterior ethmoid sinus, three cases with vascular encasement in the subdural space, and three cases in the superior sellar region.
Thirty-eight patients (48.1%) experienced surgical complications. Intracranial infection was found in 18 patients, one of which was treated by EEA, and the other 17 were operated on through open approaches. Pneumonitis was found in six patients, and two of them died of pneumonitis two and three months after the operation. Abducens nerve injury was found in seven cases, trigeminal nerve injury was found in two cases, facial nerve injury was found in five cases, and IX and X cranial nerve injuries were found in six cases. Subcutaneous effusion was found in six open approach surgery cases. Soft palate healing was delayed in two cases that underwent a transoral approach. The weakness of limbs was found in three cases. Two patients had to rely on ventilators for a short period because of brain stem injury. Three patients who underwent a far-lateral approach developed hydrocephalus and were treated by shunts. Another patient who underwent a far-lateral approach experienced CSF leakage. One patient whose tumor was located in the superior-middle clivus experienced hypopituitarism. There was a significant difference in the complication rate between endoscopic approaches (22.7%, 5/22) and open approaches (57.9%, 33/57) (P = 0.005). Interestingly, no significant difference in the rate of complications was found between the GTR and non-GTR groups (50% vs 52.4%, P > 0.05).
Twenty-four patients underwent radiotherapy (RT) as adjuvant therapy in several different institutes. Five of them died of the disease and the other nineteen patients are living. Three patients underwent RTs twice; the other 21 patients underwent RT once. These RTs included nine gamma knives, four cyber knives, four traditional photon therapies, five-proton therapies, and one heavy-particle RT. Four patients' radiation types were not specified. Among these 24 patients, one patient underwent surgery one year after gamma-knife therapy because of tumor progression, which might indicate that RT should be avoided as an initial treatment.
Except for one patient, all of the others were followed up. Follow-up ranged from 3 to 139 months (mean 66.5 months), which was calculated from the date of the first operation performed in our department to the date of the last follow-up. Forty-four patients were alive and seventeen deaths occurred. Fifteen patients died of tumor progression at 10–103 months (mean 42.2 months) after initial surgery, and two patients died of surgical complications (pneumonitis). Among the death cases, two cases showed evidence of tumor metastasis: one was thoracic metastasis and the other was neck lymphatic metastasis.
The estimated mean OS time was 139.7 months (95% CI 121–158 months). The estimated mean PFS was 73.8 months (95% CI 58.4–89.2 months). Kaplan-Meier analysis predicted OS rates of 77.9% and PFS rates of 46.1% at 5 years. The GTR group showed better survival compared with the non-GTR group (P = 0.043). All patients in the GTR group remained alive, except for one patient who died 100 months after surgery. Although the PFS was not significantly different between the GTR group and the non-GTR group (P = 0.079), there was still a trend toward better PFS in the GTR group than in the non-GTR group (99.9 ± 14.6 vs 62.8 ± 8.3 months) [Figure 5].
There was no significant difference in the OS time between the group with RT and the group without RT (P = 0.559). However, this finding does not rule out an effect of RT because the RT procedures were not standard in the series.
It is reported that PAC patients account for less than 5% of all chordoma patients. The proportion of PAC patients is 12.0% in our hospital, which is considerably higher than that reported in the surveillance, epidemiology, and end result (SEER) database. This is the most robust PAC dataset collected to date in the Chinese population, although the sample is derived from a single center. According to the SEER database, primary chordomas were more common among males in the pediatric populations, yielding a male-to-female ratio of 1.39:1. In our database of PAC patients, the number of female patients was higher than that of male patients (a male-to-female ratio of 1:1.38).
The images used in the diagnosis of chordoma are often obvious. In this series, unusual MRI manifestations are shown in [Figure 1] and [Figure 2], which were both proven to be chondroid chordoma. Hence, although rare, the differential diagnosis of pediatric and adolescent skull base lesions should include chordoma. The most common location was the craniocervical junction area, followed by the whole clivus. As mentioned above, the tumors were almost always large at diagnosis.,
Surgical treatment and outcome
Resection remains the initial choice for skull base chordomas.,, Different surgical approaches have been adopted over the years to resect skull base chordomas. In our department, chordomas in PAC patients have historically involved a craniotomy. In recent years, EEAs have been an essential technique for the resection of skull base tumors in adults.,, EEA represents a direct and minimally invasive approach to the suprasellar, retrosellar, and retroclival spaces in the midline region, with the advantage of providing safe and effective visualization and avoiding brain retraction and crossing cranial nerves., Traditional challenges in pediatric EEA, such as small working spaces and incomplete pneumatization of the sinuses, are now less important issues. EEA resection of cranial base chordomas is safe once the adequate experience is gained with the technique., To our knowledge, EEA techniques in children have only been reported as single case reports and only in the context of purely sellar lesions. EEA was adopted in 1998 at our center. In our cohort, there were 22 endoscopic surgeries (17 EEAs) for skull base chordoma. We notice a trend toward increasing numbers of EEA chordoma resections in our department each year. We encourage combining the use of endoscopy and microscopy when a lateral approach is necessary because endoscopy is helpful in finding hidden residual tumors.
The goal of chordoma resection should be GTR. In our series, OS time was longer in the GTR group than in the non-GTR group. This result confirms the efficacy of resection in improving outcomes.,,,,, Unfortunately, tumors are often diagnosed with large volumes, brainstem distortion, and the involvement of vital nerves and vasculature in PAC populations. To perform GTR with acceptable morbidity is often challenging. Those chordomas located in the lower clivus and craniocervical junction area are more challenging for neurosurgeons, even with staged surgeries. For tumors in which GTR is impossible, partial resection is also helpful for long-term control. In such situations, surgery should aim to decompress the brainstem and optic pathway and reduce tumor volume to enhance the effectiveness of subsequent RT. No significant difference in the rate of tumor resection between primary and recurrent chordomas was found in this series, which might be due to the relatively small sample size. However, we did notice a trend toward a higher resection rate in the primary chordoma group, which supports the belief that optimum resection is likely to occur during the first attempt.
The morbidity rate of complications has been reported from 22.6% to 60.3%, with a mortality rate of 2.0–7.8%. In our series, the rate of surgical complications, especially associated with open approaches, was similar to that reported in previous literature. Lateral approaches require operating in the narrow space between the brain parenchyma and skull base bone and cross cranial nerves and vasculature, which may lead to brain damage during retraction, cranial nerves palsy, and increased opportunity for infection.
Chordoma patients take RT as adjuvant therapy, a practice that has already been adopted by many medical centers., However, controversies about the effects of RT exist; some results support the use of RT after surgery, while others suggest that RT may be associated with decreased survival., With respect to PAC patients, to our knowledge, only sparse literature exists., Due to concerns about late side effects of RT, such as growth hormone deficiencies, PAC patients who received surgery alone account for a large portion (61.2%) of our sample. A similar finding was observed in the SEER database (51.2 vs 45.5%, P < 0.001). We failed to find a statistical benefit of RT. This may be related to the limited number of cases and inconsistent RT modalities and doses. However, there was still a trend indicating that patients receiving RT may have better outcomes. In another series, pediatric clival chordoma treated with combined radical resection and adjuvant proton-beam RT achieved an excellent outcome. We believe that high-dose proton therapy plays a vital role in addition to radical resection in that series. Therefore, we concur that a multidisciplinary team, including a radiation oncologist, should cooperate in the treatment of PAC patients. Further standardized RT research is necessary to provide strong evidence.
We believe that RT as an initial treatment should be avoided because we found that tumor texture is hard and, thus, total resection with microinstruments is almost impossible without morbidity. Furthermore, a recent study described histopathological parameters that correlated with prognosis, and these parameters may help us identify a subgroup of patients of low pathological risk for whom RT can be postponed or avoided. Another consideration is that the GTR rate is highest when tumor volume is small so that, even for the small tumor, we recommend surgery be chosen as the initial treatment and applying RT only after a pathology specimen is acquired.
Other prognostic factors
Although delayed local failure and a higher rate of metastasis were reported and younger age is a risk for disease progression, the long-term outcome is excellent in PAC patients compared to adults.,, We also observed an optimal OS and PFS (139.7 months and 73.8 months, respectively) in an Asian population (Chinese), although the Asian/Pacific Islander race was identified as an independent prognostic factor for increased mortality (OR = 8.4). It seems that the long-term outcome for PAC is not so discouraging. Histopathology type is an important prognostic factor. In our series, one dedifferentiated case was diagnosed. She received RT one month after surgery and died 11 months after surgery. Two patients (3.2%) showed evidence of tumor metastasis, which is clearly fewer than were observed in the SEER database (14.8%). This could be explained by the following: our follow-up was not long enough, and the different races might have different biological behaviors.
PACs account for a higher proportion (12%) of skull base chordomas than previous data had suggested. Additionally, females account for a higher proportion of cases in this population. GTR is the most effective treatment for PACs to date. The most frequent site of PACs is the craniocervical junction area; total resection is still a challenge. The role of RT in PAC populations should be further investigated with a well-designed procedure.
Declaration of patient consent
The authors certify that they have obtained all appropriate patient consent forms. In the form, the patient(s) has/have given his/her/their consent for his/her/their images and other clinical information to be reported in the journal. The patients understand that their names and initials will not be published and due efforts will be made to conceal their identity, but anonymity cannot be guaranteed.
Financial support and sponsorship
The Beijing Natural Science Foundation (No. 81771489) and the Beijing Municipal Science & Technology Commission (No. Z171100000117002).
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5]
[Table 1], [Table 2], [Table 3]