Fractionated stereotactic radiation therapy improves cranial neuropathies in patients with skull base meningiomas: a retrospective cohort study

We reviewed all patients with skull base meningiomas treated with FSRT from 1/1/1994 to 3/1/2009. Skull base meningiomas were defined as any clinically or pathologically diagnosed meningioma which involved the cavernous sinus, sphenoid, clinoid, sella, suprasellar region, cerebellopontine angle, petroclival region, foramen magnum, clivus, or designated as skull base, not otherwise specified (which tended to be large tumors encompassing multiple regions). For analysis of local control, we classified patients into four treatment categories: RT alone, adjuvant RT (after subtotal or near total resection), RT for progression/recurrence after surgery, or re-irradiation (with or without prior surgery).

All patients were treated on a dedicated stereotactic linear accelerator (Varian SR600 or Novalis TX). Gross tumor volume (GTV) was equivalent to clinical target volume (CTV) and planning target volume (PTV), and was defined as the contrast enhancing tumor only. Our standard prescription dose was 54 Gy (range 10 – 60 Gy, median 52.2 Gy) in 1.8 or 2.0 Gy/fraction. Dose was reduced in 56% of cases when the optic chiasm dose exceeded 56–57 Gy, or in re-irradiation cases. Prior to 2004, treatment planning was performed on the Radionics XKnife (Radionics, Inc., Burlington, MA) treatment planning system, with multi-isocenter plans (median 79% isodose line) using spherical collimators and Gill-Thomas-Cosman frames for patient immobilization. After 2004, patients were planned with BrainLab (BrainLAB AG, Feldkirchen, Germany) treatment planning system with single isocenter conformal plans using 5 dynamic arcs (median 90% isodose line), and thermoplastic mask for patient immobilization with position verification using ExacTrac kV daily imaging. Static stereotactic intensity modulated radiation therapy (IMRT) was introduced in 2004, and was generally used for large tumors with eccentric shapes or those close to critical structures.

Patients followed up with a radiation oncologist at 6 weeks after completion of RT. Subsequent follow-up visits with either a radiation oncologist or neurosurgeon occurred at 3 months, 6 months, and 1 year, and then yearly and bi-annually up to 10 years, and then per patient preference afterwards. Length of follow-up was defined as time from the start of RT to the last follow-up with radiation oncology or neurosurgery. Time to progression was defined as the time from start of RT to the date of first radiographic evidence of progression.

Symptoms were classified into major categories based on the cranial nerves (CN) involved: visual field/visual acuity (CN2), extra-ocular movements/ptosis (CN3, 4, 6), facial sensation (CN5), facial strength (CN7), hearing/balance/tinnitus (CN8), swallowing/tongue weakness (CN9-12), and other (including long tract signs and symptoms due to mass effect such as headache and proptosis). We included all symptoms at time of RT, including those which first appeared in the post-operatively. Duration of symptoms prior to RT was calculated from date of initial symptom presentation to start of RT. Durability of symptom improvement or worsening was defined as a continued subjective change or physical exam difference at the time of last followup compared to pre-RT baseline. Time to improvement and time to worsening were calculated from the start of radiotherapy to the date of change. Adverse events were defined as development of a new cranial neuropathy such as radiation induced optic neuritis (RION), trigeminal neuropathy, or other symptoms not associated with tumor progression. Radiographic findings of radiation necrosis were included as adverse events, regardless of symptoms.

Statistical analysis was performed using SAS v9.2 (SAS Institute Inc, Cary, NC). Freedom from progression was estimated using Kaplan-Meier method, with differences between treatment groups calculated using the log-rank test compared to the RT alone group. Events were censored at last follow-up. We used univariate and backwards stepwise multivariate Cox proportional hazards models to estimate the hazard ratios of factors associated with tumor progression. The initial multivariate model included age, sex, prior RT, midline or bilateral location, presence of symptoms at presentation, use of IMRT, use of BrainLab planning system, prior surgery, treatment group, GTV size, fraction size (≥2 Gy vs. < 2 Gy), total dose, time from diagnosis to RT, and pathologic grade, if known. A two sided α < 0.05 was considered significant.

Univariate and backwards stepwise logistic regression model was used to determine factors associated with any symptom improvement. Covariates in the initial multivariate model included duration of symptoms prior to RT, age, gender, GTV size, fraction size, total dose, prior surgery, re-irradiation, midline or bilateral location, use of IMRT, and use of BrainLab planning system. We used logistic regression to determine the odds ratio (OR) of worsening of symptoms events in cases with progression compared to those without progression. A two sided α <0.05 was considered significant.

This retrospective study was conducted after approval by the institutional review board at Thomas Jefferson University. We identified 278 patients treated with FSRT for skull base meningiomas from 1/1/1994 through 3/1/2009. Follow-up data were available on 217 patients (78.1%). The median follow-up time was 4.4 years. Seven patients were treated with a second course of FSRT, totaling 224 courses of FSRT included in the analysis. We considered each treatment course as a separate case for the purposes of analyses (Table 1). The majority of patients were female (74%), and the median age at the time of radiation therapy was 60.7 years old. Ninety-two percent of cases were symptomatic at the time of FSRT; the remaining cases were treated for asymptomatic progression. The most common area of involvement was the cavernous sinus (30%) or sphenoid/clinoid region (29%). These data represent approximations as tumors often encompassed multiple regions. Treatment consisted of RT alone in 51% of cases, adjuvant RT in 18%, RT for progression/recurrence after prior surgery in 26%, and re-irradiation in 5% of cases.

Twenty-one patients progressed after FSRT, representing a 90.6% cumulative freedom from local failure. Local control at 5 years was 96.1% in the RT alone group, 96.8% (p=0.75) in the post-operative group, 77.3% (p=0.04) in the progression/recurrence after surgery group, and 44.4% (p < 0.01) in the progression after radiation group (Figure 1). In addition to treatment group, recurrence after prior RT (Hazard ratio (HR) 8.4, p<0.01), midline/bilateral tumor location (HR 2.9, p=0.02), any prior surgery (HR 2.51, p=0.05), treatment volume per each cm [3] larger (HR 1.032, p=0.03) and use of IMRT (HR 2.73, p=0.04), were significantly correlated with increased rate of local failure on univariate analysis. Total dose, use of BrainLab planning system, being symptomatic at presentation and fraction size were not correlated with local failure. On multivariate analysis, midline/bilateral location (HR 2.97, p=0.02), atypical histology (HR 8.70, p=0.01) and treatment group were associated with local failure. Compared to RT alone, adjuvant RT (HR 0.82, p=0.79) and RT for recurrence after surgery (HR 1.78, p=0.32) were not significantly different; however, re-irradiation (HR 11.02, p<0.01) was a significant predictor of subsequent local failure.

Patients were symptomatic at the start of treatment in 92% of cases. The most common presenting symptoms were decreased visual acuity/visual field deficits (57.6%) and diplopia/ptosis (33.5%) (Table 2). In our cohort, the frequency of patients who received surgery prior to RT differed in each symptom category. Approximately 40% of cases presenting with diplopia or ptosis had surgery prior to RT, while 90% of those presenting with facial weakness or CN9-12 deficits had surgery prior to RT. Symptoms were first noted in the post-operative setting in several cases, including facial weakness (50%), CN9-12 deficits (37.5%) and decreased facial sensation (18%). The median duration of symptoms prior to radiation was generally around 1 year (Table 2), and was longer in cases with prior surgery (data not shown).

Symptom outcomes varied according to initial symptoms (Figure 2). The outcomes of 10-30% of symptoms after FSRT were not documented, particularly for those related to hearing and balance. Among cases without tumor progression, 30-40% had durable improvement, 30-40% had stable symptoms, and less than 10% had symptom deterioration after FSRT (Figure 2). The most common presenting symptom, visual acuity and visual field deficits, was durably improved in 37% of cases, maintained in 35%, and deteriorated after tumor progression or treatment related adverse event in 14%. Additionally several patients developed new symptoms attributable to co-morbid conditions, such as worsening vision in a patient with HSV keratitis or new deficits following a stroke. Durable symptom improvement of at least one symptom occurred in 57% of cases.

Symptom improvement typically occurred within 6 months of the start of FSRT (Table 3), and often earlier. By the last day of FSRT course, 17% of cases had experienced improvement in at least one symptom, and by 2 months after FSRT, 27% of cases had improvement in at least one symptom. Of cases with symptom improvement, 33% first occurred by the end of FSRT and 54% within 2 months after FSRT. In contrast, symptom deterioration was observed many months later (median 18.7 months). In the absence of tumor progression, 89% – 100% of symptom improvements were durable at the time of last follow-up. For an illustrative example of early symptom improvement (Table 4 and Figure 3).

Symptom worsening occurred more often in cases where tumors progressed (68% vs. 13%, p < 0.01). Patients with tumor progression had a 12 fold increased odds of symptoms worsening (OR 12.29 95% CI 4.6 – 32.6, p < 0.01). The most common presenting symptom, visual field/visual acuity deficit, worsened in 12 of 17 (70.6%) patients who ultimately experienced progression. Worsening of symptoms commonly preceded objective radiographic finding of tumor progression (data not shown).

On univariate analysis, prior surgery (OR 0.44, 95% CI 0.25-0.78), midline/bilateral location of tumor (OR 0.44, 95% CI 0.22-0.90), and re-irradiation (OR 0.18, 95% CI 0.38-0.88) were negative predictors of symptomatic improvement after FSRT. Of note, GTV size, fraction size, total dose, use of IMRT, use of BrainLab planning system, and duration of symptoms prior to FSRT were not associated with symptom improvement. On multivariate analysis, any prior surgery (OR 0.47, 95% CI 0.25-0.86), and age (per 10 years older) (OR 0.78, 95% CI 0.62-0.98) were negative predictors of symptom improvement, while re-irradiation (OR 0.40, 95% CI 0.11 – 1.40), midline/bilateral location was no longer significant (OR 0.53, 95% CI 0.25-1.11) (Table 5).

We noted 32 adverse events in 28 patients (12.5% cumulative incidence). The most common adverse events were radiation induced optic neuritis (n = 5, 2.2%), radiation necrosis (n = 5, 2.2%) and facila numbness or pain (n = 5, 2.2%). Additional adverse events included three cases of hypothyroidism, two cases each of diplopia, significant cognitive decline, and occipital neuralgia, and one case each of dysphagia, facial myoclonus, glossopharyngeal neuralgia, seizures,, diabetes insipitus, and vertigo.