The impact of facility type and volume on treatment and overall survival in craniopharyngioma

Craniopharyngiomas are uncommon benign sellar and parasellar tumors with high overall survival (OS) and recurrence rates. Treatment is often surgical but may include adjuvant therapies. The impact of adjuvant therapy and surgical approach have been evaluated, however, facility volume and type have not. The purpose of this study is to analyze the influence of facility volume and type on treatment modalities, extent of surgery and survival of craniopharyngioma. The 2004–2016 National Cancer Database (NCDB) was queried for patients diagnosed with craniopharyngioma. Facilities were classified by type (academic vs. non-academic) and low-volume center (LVC) (Treating < 8 patients over the timeline) versus high-volume center (HVC), (Treating ≥ 8 patients over the timeline). Differences in treatment course, outcomes, and OS by facility type were assessed. 3730 patients (51.3% female) with mean age 41.2 ± 22.0 were included with a 5-year estimated OS of 94.8% (94.0–95.5%). 2564 (68.7%) patients were treated at HVC, of which 2142 (83.5%) were treated at academic facilities. Patients treated at HVC’s were more likely to undergo both surgery and radiation. Surgical approach at HVC was more likely to be endoscopic. Patients treated at HVC demonstrated significantly higher 5-year OS compared to patients treated at LVC (96% [95% CI 95.6–97.1% versus 91.2% [95% CI 89–92.7%] with lower risk of mortality (Hazard ratio [95% CI] = 0.69 [0.56–0.84]). Treatment of craniopharyngioma at HVC compared to LVC is associated with improved OS, lower 30- and 90-day postoperative mortality risk, and more common use of both radiotherapy and endoscopic surgical approach.


Introduction
Craniopharyngioma is an uncommon benign sellar and parasellar tumor that arises from the craniopharyngeal duct and occurs in both the pediatric and adult populations [1].Incidence has been reported to be between 0.5 and 2 cases per million persons per year with approximately 338 new cases each year in the United States [2,3].Histological subtypes include adamantinomatous (ACP) and papillary (PCP) craniopharyngioma with a bimodal peak of incidence for ACP (5-15 years and 45-60 years) and peak incidence in the fifth and sixth decades of life for PCP [1].These patients often present with severe/chronic headache, vision loss, and/ or endocrinopathies given the anatomical proximity of the tumor to the optic chiasm and pituitary.Multiple long-term population-based studies have demonstrated a relatively high overall survival (OS) with 5-year OS rates greater than 80% in adults and over 95% in children [2,[4][5][6][7][8][9].Although rare, patients may ultimately succumb to death due to circulatory and respiratory disease secondary to hypopituitarism [8,9].
Despite the high OS, the optimal management of craniopharyngioma remains a debate.as treatment protocols vary depending on patient tumor characteristics while attempting to balance disease control and treatment-related toxicity [10].Craniopharyngiomas are most often treated by surgery A portion of this work was presented as a podium presentation for the North American Skull Base Society at the NASBS 31st Annual Meeting, February 19, 2022.alone, however, nearly 1 in 4 cases are also treated with adjuvant radiation therapy (RT) or hormone therapy [5].The impact of surgical approach, extent of resection, and adjuvant therapy on OS of craniopharyngioma have been evaluated and found to be noncontributory despite improved local control [5,[10][11][12].Outcomes and treatment of head and neck and skull base tumors have been evaluated in relation to facility volume and type in previous studies including laryngeal cancer, pituitary adenoma, nasopharyngeal carcinoma, esthesioneuroblastoma, sinonasal squamous cell carcinoma, among others [13][14][15][16][17][18][19][20][21][22].However, the influence of facility volume and type on treatment and survival of craniopharyngioma have yet to be determined.Given the complexity and multidisciplinary care needed for management, this study investigates a large hospital-based database, the National Cancer DataBase (NCDB), to determine the influence of facility volume on treatment modalities and survival of craniopharyngioma.

Study population
This was a retrospective cohort analysis of the National Cancer DataBase (NCDB), a publicly available, deidentified clinical oncology database which collects case data from over 1500 facilities accredited by the Commission on Cancer.We queried the NCDB from 2004 to 2017 using the International Classification of Diseases for Oncology, Third Edition (ICD-O-3) histology codes 9350-9352 for craniopharyngioma.Exclusion criteria included patients receiving palliative care, treatment occurring at facilities separate from the reporting facility, and patients with more than one primary tumor.Cases with missing survival data were excluded.Because the NCDB is a publicly available, de-identified database, the study was exempt from review by the University of California, Irvine Institutional Review Board.

Study variables
Facility volume was the primary independent variable and was defined as a categorical variable.High volume was defined as greater than the 80 th percentile of total cases reported during the study period, which was 8 cases.Secondary independent variables included age, which was binarized as 0-18 years or > 18 years to represent the bimodal age distribution of craniopharyngioma.The Charlson-Deyo Comorbidity Index was also binarized as 0 or ≥ 1 as few patients had an index of 2 or 3. Tumor size was defined as either < 3.2 cm or ≥ 3.2 cm, which was the median tumor size of our sample.Facility type was defined as non-academic (Community Cancer Programs, Comprehensive Community Cancer Programs, and Integrated Network Cancer Programs) or academic (includes NCI-designated comprehensive cancer centers).Insurance type was defined as private, government (Medicaid, Medicare, or Other Government), not insured, or unknown.Treatment modality was defined as surgery alone, radiation alone, or surgery and radiation.
Other secondary variables included sex, race, distance to treatment facility, urban-rural residence, histology, and facility region.For patients who received surgery, surgical approach and extent of surgery were included as secondary variables.

Outcomes
The primary outcome of this study was OS.Additional analysis of surgical outcomes was performed.These secondary outcomes included time from diagnosis to surgery, postoperative length of stay, 30-and 90-day post-operative mortality, and unplanned 30-day readmissions.

Statistical methods
Statistical analysis was performed using SAS 9.4.Categorical variables were compared using Chi-squared testing.Continuous variables were compared using two-tailed t-testing, or alternatively, Wilcoxon Rank-Sum testing for variables with significantly non-normal distributions.Kaplan-Meier log-rank testing was used to estimate survival times.Univariate Cox proportional hazards analysis was used to analyze the association between independent variables and overall survival.Variables which were statistically significant (p < 0.05) were then entered into a multivariate Cox proportional hazard analysis of facility volume and overall survival.

Results
A total of 3730 patients (51.3% female) with craniopharyngioma were included in the study cohort.The mean age was 41.2 ± 22.0 years and demonstrated a bimodal distribution with peaks at approximately 10 and 54 years.The majority of patients were White (64.0%) and had private insurance (56.7%).Patients were treated at 602 unique facilities, which treated as many as 70 cases during the study period.The high-volume facilities accounted for 2564 (68.7%) cases included in the study.The 1-, 2-, and 5-year estimated OS for the study cohort was 98.4% (95% confidence interval [CI] 97.9-98.7%),97.2% (95% CI 96.6-97.7%),and 94.8% (94.0-95.5%),respectively.
Several statistically significant differences exist between patients treated at high-volume (HVC) vs lowvolume (LVC) centers (Table 1).Patients treated at HVC were more likely to be younger, have a Charlson-Deyo comorbidity index of 0, have private insurance, and to live further from the treatment facility.These patients were also more likely to be treated at academic facilities in the Northeast, and to undergo treatment with both surgery and radiation.Surgical approach at HVC was also more likely to be endoscopic.
Kaplan-Meier log-rank analysis demonstrated that patients treated at HVC had a higher 5-year OS compared to patients treated at LVC (96% [95% CI: 95.6-97.1% versus 91.2% [95% CI 89-92.7%],p-value < 0.001).Survival curves are presented in Fig. 1.Univariate Cox proportional-hazards analysis identified several statistically significant covariates (Table 1).LVC, age > 18 years, Charlson-Deyo Comorbidity Index ≥ 1, treatment at a facility in the South, Black race, tumors ≥ 3.2 cm, treatment with both surgery and radiation, and open or unspecified surgical approaches were associated with a higher risk of mortality.Conversely, HVC, increasing distance from the treatment facility, academic facilities, private insurance, and treatment with radiation alone were associated with a lower risk of mortality.
After controlling for covariates, multivariate Cox proportional-hazards analysis demonstrated that treatment at HVC was significantly associated with a lower risk of mortality (Hazard ratio [95% CI] = 0.69 [0.56-0.84]).Other covariates which remained significant included age, Charlson-Deyo comorbidity index, insurance types, tumor size, and treatment modality (Table 2).Facility type was not included as a covariate due to significant collinearity with facility volume.Surgical approach was not included due to both a large percentage of patients not receiving surgery and a substantial proportion of missing data.
Secondary analysis was performed using a subgroup of 2942 patients who were treated surgically, with or without radiation (Table 3).Patients undergoing surgery at HVC were more likely to receive radiation (22.9% vs 14.1%, p < 0.001) and experienced longer time intervals between diagnosis and surgery (24.0 vs 17.1 days, p < 0.001).While there was no statistically significant difference between postoperative LOS and unplanned readmissions, patients treated surgically at HVC had lower rates of both 30-and 90-day mortality (1.5% vs 4.2%, p < 0.001 and 3.0% vs 7.2%, p < 0.001, respectively) (Table 4).

Discussion
This study is the first to describe the influence of facility type and volume on treatment modalities, extent of surgery, and OS of craniopharyngioma using a national database.Patient sociodemographic factors associated with treatment at HVC included patients who were younger, had private insurance, and lived farther from the treating center.Patients presenting to academic centers were also more likely to have private insurance and travel longer distances for treatment.
There was a significant difference in treatment modalities employed, with patients at high volume facilities being more likely to receive adjuvant radiotherapy.Additionally, we observed a statistically significant increase in OS for patients treated at academic facilities compared to non-academic facilities and HVC compared to LVC.Upon further analysis, mortality was lower for patients treated at academic centers and HVC compared to non-academic facilities and LVC.These findings indicate that the lower OS and higher mortality associated with LVC and non-academic centers may be a source of disparity in the treatment of craniopharyngioma in patients who do not have the ability to obtain treatment at HVC and academic centers.Though it may not always be the case, academic centers are likely to encounter a higher volume of craniopharyngioma cases due to the complexity of their treatment, which typically includes surgical resection with the addition of chemotherapy and/or radiation [7,23,24].This is due in part to the invasive nature of the tumor with involvement of nearby critical neurovascular and neuroendocrine structures.Surgical and adjuvant treatment planning is paramount for successful control of craniopharyngioma requiring collaboration by an expert multidisciplinary team and resources to treat these complex skull base lesions [7,10,[25][26][27].Treatment directed by a multidisciplinary team is associated with improved OS for multiple sinonasal and skull base tumors [25][26][27].These teams are typically available at academic centers with a high-volume of complex cases and include experienced neurosurgeons, otolaryngologists, endocrinologists, neuro-oncologists, radiation oncologists, and neuroradiologists [16].Previous studies have demonstrated craniopharyngiomas are more often treated at academic centers, however, the influence of facility type and volume on treatment and overall survival of craniopharyngioma was previously not reported [10,28].
Surgical resection of craniopharyngiomas can be performed endoscopically or non-endoscopically.It has been shown that endoscopic resection is associated with lower mortality, fewer complications, and shorter hospital length of stay than non-endoscopic resection for craniopharyngioma, and that academic facilities are more likely to utilize endoscopic technique [28,29].This study found no difference in the extent of surgery performed at HVCs; however, it was found that HVCs were more likely to utilize endoscopic technique, which may contribute their higher OS.This finding was not controlled for covariates and therefore may be confounded by other factors, however this is not likely given that larger tumors (> 3cm) and those requiring adjuvant radiotherapy have been shown to be more likely to require a non-endoscopic approach, and according to this study, both of these are characteristics were associated with HVC.Larger craniopharyngiomas, which were more likely to be treated at HVC, may require a transcranial approach, which is associated with a higher number of adverse events [30,31].Higher OS and lower mortality despite larger tumor size may be due to multidisciplinary approach as previously mentioned, or due to increased surgeon experience [32].
Radical tumor resection is associated with higher recurrence-free survival, however subtotal resections provide lower risk to surrounding structures, and is often combined with radiotherapy [31,33].Adjuvant radiotherapy can be used to reduce recurrence or treat residual tumor following surgery depending on tumor size, location, and extension, as well as the risk for local toxicity from radiation [33].Recently, gross total resection (GTR) and subtotal resection with adjuvant radiation have been shown to improve OS [24].It has been shown previously in cases of head and neck malignancy that patients at HVC and academic facilities receive more aggressive treatment, and are also more likely to be treated according to evidence-based treatment guidelines [14,22,34].This study found that those at HVC were more likely to have received radiotherapy in addition to surgery; HVC were shown to encounter tumors of larger size, which makes total resection more difficult and may have contributed to this increased use of adjunct radiotherapy.Radiotherapy does introduce the risk for acute and chronic toxicity, which may lead to visual impairment and endocrine deficiency [35].Newer radiation technologies like intensity-modulated photon therapy and proton-beam radiation therapy may lead to fewer off-target adverse effects due to decreased radiation to surrounding tissues [33].Though this study does not stratify by radiation type, it could be hypothesized that academic and HVC would utilize these newer technologies at higher rates, resulting in better outcomes, though further studies would be needed to support this claim.
Another patient characteristic associated with care at HVC's was younger age.Given that the already high level of complexity of craniopharyngioma surgery is intensified by the still-developing and highly sensitive neuroendocrine structures in the pediatric population, as well as the fact that younger patients have been shown to present with larger sized tumors than adults, it is logical that younger patients seek care where a team of multidisciplinary specialists is easily accessible [10,33].Additionally, approximately 33-35% of craniopharyngioma patients are children, who typically have fewer pre-existing comorbidities than adults [36].This may contribute to the lower Charlson-Devo comorbidity index associated with HVC found in this study.This also brings up the issue that older patients with increased comorbidities are at higher risk for all-cause mortality, which may be a partial contributor to the increased mortality at LVC.
Lower socioeconomic status often present with larger, more advanced tumors, as uninsured patients have been shown previously to present with more advanced disease [37].This was not observed in this study, likely owing to craniopharyngiomas' slow-growing nature, which contributes to late diagnosis across patient populations [38].However, patients at HVC and academic centers were more likely to have private insurance, supporting the idea that care at these facilities may be less accessible to those with lower socioeconomic status.Another source of disparity revealed by this study is that patients at HVC and academic centers were more likely to live farther from the treatment facility.The ability to travel long distances for treatment requires reliable transportation and perhaps the ability to arrange for more time off work.Additionally, patients who have a higher degree of disability, whether related or unrelated to their tumor, may not have the ability to travel long distances, especially for adjuvant radiotherapy, which requires repeated visits [39].Previous literature has demonstrated similar findings with regard to insurance status and distance traveled in other head and neck tumors as well [23,40].
Lastly, patients at HVC experienced longer time intervals between diagnosis and surgery, which is logical owing to the higher volume of cases.In highly aggressive malignancies, the increased surgical delay associated with HVC and academic centers has been shown to correlate with increased mortality [23].However, craniopharyngiomas are histologically benign and slow-growing, with overall low mortality, so longer surgical wait times appear to be a worthwhile risk given the benefit of lower 30-and 90-day mortality that our study has shown [41,42].

Limitations
Common to most database studies, this study contains several limitations worth mentioning.First, the NCDB is a de-identified national database which may result in inaccurate or missing information, inclusion bias, and limited number of variables.Craniopharyngiomas have high recurrence rates, and despite high survival rates overall, almost all patients will develop some sequelae as a result of craniopharyngioma and/or tumor resection [38,43].Unfortunately, due to the nature of the NCDB reporting, we are unable to analyze disease-free survival and recurrence rates, which may further explain benefit of treatment at academic facilities and HVC.The NCDB also does not report intraoperative outcomes, which depending on surgical approach, may include CSF leak as well as damage to neuroendocrine or vascular structures near the craniopharyngioma during surgery [44].Additionally, there is no detailed information available regarding tumor characteristics other than size (such as extent of tumor invasion), and GTR classification is based on surgeon impression rather than any quantitative measure.
In NCDB, open approaches include both craniotomies as well as microsurgeries; therefore, these surgical approaches cannot be assessed separately.Lastly, in multisurgeon operations, which are common in multidisciplinary cases such as these, the outcomes of otolaryngology co-surgeons are not able to be analyzed as NCDB does not specify which surgical specialties participate in each case.

Conclusion
HVC offer improved OS, lower 30-and 90-day postoperative mortality risk, and more common use of both radiotherapy and endoscopic surgical approach when compared to LVC.These patients more often travelled greater distances and were privately insured suggesting access to HVC may be limited for patients with fewer socioeconomic advantages.Future work may focus on differences in recurrence and morbidity by hospital volume and type, as well as strategies to mitigate the inequity in care that patients with lower socioeconomic status experience when seeking treatment for craniopharyngioma.

Table 1
Facility, demographic, and clinical characteristics for patients treated at High-Volume versus Low-Volume Centers Low Volume High Volume P-value

Table 3 (
Author contributions BFB: Conceptualization, methodology, writing-original drafting, writing-review & editing, approval of final version.DHL: Conceptualization, methodology, statistical analysis, writing-original drafting, writing-review & editing, approval of final version.MMN: Conceptualization, writing-original drafting, writing-review & editing, approval of final version.SJT: Conceptualization, methodology, writing-original drafting, writing-review & editing, approval of final version.FPKH: Scientificc advisor, approval of final version.ECK: Conceptualization, methodology, writingoriginal drafting, writing-review & editing, supervision, approval of final version.Funding The authors did not receive support from any organization for the submitted work.