Modern LINAC-based radiotherapy is safe and effective in the treatment of secretory and non-secretory pituitary adenomas

Abstract

Objective:

Adjuvant radiotherapy (RT) can help achieve local tumor control (LC) and reduce hormonal overexpression for pituitary adenomas (PAs). Prior reports involved Gamma Knife or older LINAC techniques. We report on long-term outcomes for modern LINAC RT.

Methods:

Institutional retrospective review of LINAC RT for PAs with minimum 3 years MRI follow-up. Hormonal control defined as biochemical remission in absence of medications targeting hormone excess LC defined using RECIST on surveillance MRIs. Progression Free Survival (PFS) defined as time alive with LC and without return of or worsening hormonal excess from secretory PA. Kaplan-Meier method and Cox proportional hazard models used.

Results:

From 2003-2017, 140 patients with PAs (94 non-secretory, 46 secretory) were treated with LINAC RT (105 fractionated, 35 radiosurgery) with mFU of 5.35 years. Techniques included fixed gantry IMRT (51.4%), DCA (9.3%), and VMAT (39.3%). PFS at 5-years was 95.3% for secretory tumors and 94.8% for non-secretory tumors. Worse PFS associated with larger PTV on MVA (HR 2.87, 95% CI 1.01 – 8.21, p=0.049). Hormonal control at 5 years was 50.0% and associated with higher dose to the tumor (HR 1.05, 95% CI 1.02-1.09, p=0.005) and number of surgeries (HR 1.74, 95% CI 1.05-2.89, p=0.032). Patients requiring any pituitary hormone replacement increased from 57.9% to 70.0% after radiotherapy.

Conclusion:

Modern LINAC RT for patients with PAs was safe and effective for hormonal control and LC. Notably, no difference in LC was noted for functional versus non-functional tumors possibly due to higher total dose and daily image guidance.

Introduction:

Pituitary adenomas (PAs) are benign tumors that can cause symptoms from compression on adjacent structures such as the optic apparatus, cranial nerves in the cavernous sinus, or the infundibulum, or through hypersecretory syndromes from tumors that produce excess amounts of one or more of the pituitary hormones. With an incidence of 3.9-7.4 per 100,000 per year,^1-4 PAs are predominantly treated with surgery as primary therapy for all types except for prolactinomas, where medical therapy with dopamine agonists is usually preferred.⁵ Adjuvant radiotherapy has been shown to achieve both local tumor control and reduce hormonal overexpression when surgery or medical therapy are unable to accomplish these goals. Much of the prior literature on outcomes of pituitary radiation pertains to Gamma Knife radiosurgery (GKRS),^6-10 with less information published regarding linear accelerator (LINAC)-based fractionated radiotherapy^11-13 or proton radiotherapy.^14,15 While GKRS and proton radiotherapy have promising data and, in the case of the former, is an accepted standard of care, LINACs are ubiquitous across radiation oncology centers with far fewer centers employing GKRS or proton radiotherapy. Furthermore, for tumors that abut optic nerves or the optic chiasm, single fraction radiosurgery is not feasible making LINAC-based treatment the primary option. In addition, previous reports suggest that functional PAs had significantly worse local tumor control than non-functional tumors after radiotherapy.^12,16 Here, we report our institutional results of LINAC-based radiotherapy for patients with functional and non-functional PAs.

Materials and Methods:

We performed a retrospective chart review of consecutive patients who received LINAC-based radiotherapy for histologically confirmed PA or biochemical and radiographic evidence of functional PA without resection at our institution from 2000-2018. Data collection was approved by the institutional IRB. Patient consent was waived as identifiable patient information was not recorded for analysis. Patients censored before 3 years of MRI follow-up were excluded from analysis. Patient characteristics were reported using frequencies and percentages for categorical variables and mean with standard deviation or median and IQR, as appropriate for numerical variables. Variables that did not follow Gaussian distribution were log transformed.

Progression free survival (PFS) was defined from date of radiotherapy completion to local progression, hormonal recurrence, or death with censor date at last follow-up. PFS was measured using the Kaplan-Meier method and differences between secretory and non-secretory groups were estimated using the log rank test. Median follow-up was estimated using the reverse Kaplan Meier method. Local control (LC) was defined as stable disease, partial response, or complete response of pituitary tumor based on serial MRI scans using RECIST criteria (version 1.1). Local failure, defined as progressive disease on MRI scan using RECIST criteria, was measured using the Kaplan-Meier method and the differences between the secretory and non-secretory groups was estimated using the log rank tests. Hormonal control was defined as normal pituitary hormonal parameters in absence of medications targeting hormonal excess. Date of discontinuing medications targeting hormonal excess used as event date for hormonal control. Extent of resection was defined with Gross Total Resection for patients with operative note and post-operative MRI consistent with no residual gross tumor while Subtotal resection was defined as patients who had either post-operative MRI consistent with definitive evidence of residual tumor or operative note detailing tumor left behind at end of procedure (e.g., unresectable disease in cavernous sinus). Knosp grading was based on pre-operative MRI (when applicable) or radiation planning MRI and dichotomized into Grades 0-1 and 2-4 for analysis.

Univariate cox proportional hazard model was used to test each clinicopathological variable with PFS using the firth method. Variables that were significant at an alpha of 0.2 were selected for multivariable analysis (MVA) including pituitary apoplexy and log of planning target volume (PTV) using a backward selection method. Similar univariate analysis was done within the secretory cohort using hormonal control as the event of interest. Cumulative incidence of hormonal control was estimated and difference between time to hormonal control based on receipt of SRS compared to fractionated radiotherapy was estimated using ANOVA. Factors associated with hormone replacement prior to radiotherapy were estimated using ANOVA for categorical and Pearson correlation for numeric covariates. PTV was dichotomized as < 10 cc vs ≥ 10 cc for this analysis. Those with alpha of 0.2 were selected for multiple linear regression analysis. Change in hormone replacement outcome was coded as binary (change vs no change). For the univariate analysis (UVA), chi-square, or fisher’s exact test for used to test the association with categorical variables and ANOVA for numeric covariates. Multivariable logistic regression was used with variables that met alpha of 0.1 for assessment of factors associated with change in hormonal replacement.

Results:

Demographics and Treatment Characteristics

From October 2003 to January 2017, 140 patients with PAs were treated with LINAC-based radiotherapy with median follow-up of 5.4 years (IQR 4.0-8.5). Preoperative clinical diagnosis consisted of the following: 94 non-functional, 23 growth hormone-secreting (acromegaly), 13 ACTH-secreting (Cushing’s disease), 9 prolactin-secreting, and 1 TSH-secreting PAs. One hundred and five patients were treated with LINAC-based fractionated radiation therapy and 35 patients were treated with LINAC-based single fraction stereotactic radiosurgery (SRS). Five patients with functional tumors never had any surgery for their PA, and only had medications or radiotherapy. For the remaining 135 patients, 43 had radiotherapy within 6 months of their last surgery while 91 patients had radiation greater than 6 months after surgery (median 14.5 months, IQR 5.1-38.3 months). Median residual tumor after surgery was 1.5 cc (IQR 0.4-4.9 cc) with median MIB-1 <3% (range <3% - 15%) of patients with available MIB-1 reported (n=49). Knosp grade was available for 133 patients with median Grade 1 (IQR 0-2). Patient characteristics summarized in Table 1.

Table 1.

Patient baseline characteristics.

Variable	Category	n = 140
Mean age, years (SD)		46.5 (14.5)
Sex	Male	68 (48.6%)
	Female	72 (51.4%)
Karnofsky Performance Status	100	46 (32.9%)
	90	67 (47.9%)
	80	21 (15.0%)
	70	2 (1.4%)
	60	4 (2.9%)
Pituitary Apoplexy prior to radiation	Yes	8 (5.7%)
	No	132 (94.3%)
Secretory Status	secretory	46 (32.9%)
	non-secretory	94 (67.1%)
Mean number of surgeries (SD)		1.4 (0.7)
Surgical Approach	None	5
	Transsphenoidal	128 (94.8%)
	Craniotomy or Other	7 (5.2%)
Knosp Grade	Unavailable	7
	0	36 (27.1%)
	1	36 (27.1%)
	2	28 (21.1%)
	3	10 (7.5%)
	4	23 (17.3%)
Extent of Surgery	Gross Total Resection	25 (17.9%)
	Subtotal Resection	109 (77.9%)
	Biopsy or None	6 (4.3%)
MIB-1	Unavailable	91
	<3%	31 (63.3%)
	≥ 3%	18 (36.7%)
GTV	Resection Bed Only	6
	< 4 cc	78 (58.2%)
	≥ 4 cc	56 (41.8%)
Median PTV volume (range)		14.3 (0.27- 247.60)
Type of Radiation	Fractionated	105 (75.0%)
	SRS (1-5 Fractions)	35 (25.0%)

GTV is gross tumor volume. SRS is stereotactic radiosurgery. PTV is planning target volume.

Patients were simulated with aquaplast brain mask or stereotactic head frame for immobilization with diagnostic MRI brain co-registered with treatment planning CT scan. Patients were offered SRS for tumors where the minimum distance from tumor edge to bilateral optic nerves or optic chiasm was greater than 4 mm, otherwise they were treated with fractionated radiotherapy. Fractionated radiation was delivered in 1.8 Gy per fraction to the tumor with a 1-3 mm PTV expansion, while SRS was delivered in a single fraction to the tumor with a 0-1 mm PTV expansion. Of the 140 patients, 72 (51.4%) were treated with fixed gantry Intensity Modulated Radiation Therapy (IMRT) using 6-9 gantry angles (n=3), 10 gantry angles (n=4), 11 gantry angles (n=23), 12 gantry angles (n=33), 13 gantry angles (n=4), or 14-17 gantry angles (n=5). Dynamic Conformal Arcs (DCA) were used for 13 patients (9.3%) using 1 arc (n=1), 2 arcs (n=1), 4 arcs (n=10), or 5 arcs (n=1). Volumetric Modulated Arc Therapy (VMAT) was used for the remaining 55 patients (39.3%) using 1-2 arcs (n=3), 3 arcs (n=48), 4 arcs (n=3), or 6 arcs (n=1). Non-functional PAs were treated to median fractionated dose of 50.4 Gy (IQR 50.4-50.4 Gy) and median SRS dose of 15 Gy (IQR 14-15 Gy), while functional PAs were treated to median fractionated dose of 50.4 Gy (IQR 50.4-54 Gy) and median SRS dose of 15.5 Gy (IQR 15-18 Gy). Typical fractionated constraints included max dose to 0.03 cubic centimeters (cc) of 58 Gy and volume of brainstem receiving 54 Gy to be less than 30% and max dose to 0.03 cc of optic nerves or chiasm of 54 Gy for non-functional and 55 Gy for functional adenomas. Daily image guidance was used to assure setup accuracy.

Local and hormonal control: high rates with significant time to complete hormonal control

There was no difference in PFS between secretory and non-secretory groups (log rank p=0.70). The 5-year and 10-year PFS rate was respectively 95.3% and 95.3% for secretory tumors (n=46) and 94.8% and 81.3% for non-secretory tumors (n=94) shown in Figure 1A. For patients with secretory tumors, the 5-year and 10-year PFS rate with SRS was 93.8% and 93.8% while with fractionated radiotherapy it was 96.3% and 96.3%, respectively. For patients with non-secretory tumors, the 5-year and 10-year PFS rate with SRS was 100.0% and 100.0% while with fractionated radiotherapy it was 93.5% and 74.8%. On UVA, there was no association of PFS with age, sex, KPS, secretory vs non-secretory, low vs high MIB-1, low vs high Knosp Grade, number of surgeries, surgical approach, extent of surgery, use of SRS vs fractionated radiotherapy, or EQD2 dose for alpha/beta of 3. Decreased PFS was significantly increased on UVA with pituitary apoplexy and larger log of PTV (Table 2). On MVA, only log of PTV (HR 2.60, 95% CI 1.08 – 6.28, p=0.03) was associated with PFS at a statistically significant level with weaker association with pituitary apoplexy (HR 8.33, 95% CI 0.86-100, p = 0.07) that no longer met our significance threshold. Three local failures were identified yielding an actuarial 10-year LC rate of 98.3% (Figure 1). Long term hormonal control by 5 years was 50.0% with hormonal control over time shown in Figure 2. Estimated cumulative incidence of long-term hormonal control was associated with higher 2 Gy equivalent dose to the tumor (HR 1.05, 95% CI 1.02-1.09, p<0.01) and number of surgeries (HR 1.74, 95% CI 1.05-2.89, p=0.03). There was no association on UVA between long term hormonal control and age (HR 0.99, 95% CI 0.96-1.01, p=0.25), male sex (HR 0.68, 95% CI 0.30-1.53, p=0.35), GTV size (HR 0.70 for >= 4cc, 95% CI 0.28-1.74, p=0.44), MIB-1 of ≤3% (HR 0.61, 95% CI 0.14-2.61, p=0.51), Knosp Grade 2-4 (HR 1.98, 95% CI 0.87-4.54, p=0.11), or use of 1-5 fraction radiotherapy (SRS) as compared to ≥ 6 fractions of radiotherapy (HR 1.52, 95% CI 0.74-3.14, p=0.26). MVA was not performed for secretory tumors due to limited sample size and events. Finally, similar times to hormonal control were found in patients with secretory tumors treated with either SRS (mean 3.23 years) or fractionated radiotherapy (mean 3.82 years, p=0.55).

Figure 1.

Progression free survival (PFS) of pituitary adenomas after radiation therapy by secretory status (A) and Local Control (LC) of pituitary adenomas after radiation therapy (B).

Table 2.

Univariate and Multivariable Analysis associations with Progression Free Survival.

	Univariate Analysis		Multivariable Analysis
Covariate	Hazard Ratio (95% CI)	HR P-value	Hazard Ratio (95% CI)	HR P-value
Age	1.00 (0.96-1.06)	0.898
Sex, Male vs Female	0.52 (0.09-2.16)	0.424
Karnofsky Performance Status	0.99 (0.91-1.09)	0.785
Non-Secretory vs secretory	1.22 (0.29-6.78)	0.810
Number of Surgeries	1.20 (0.39-2.98)	0.671
Type of surgery: TSA vs craniotomy/other	0.26 (0.05-2.50)	0.165
Knosp Grade: 0-1 vs 2-4	0.73 (0.16-3.04)	0.680
Extent of Surgery: GTR	0.34 (0.00-65.18)	0.631
STR	1.11 (0.12-148.15)	0.952
Biopsy or none	-	-
MIB-1: <3% vs ≥3%	0.35 (0.03-2.65)	0.382
Pituitary Apoplexy (no vs yes)	0.11 (0.02-1.12)	0.030	0.12 (0.01-1.15)	0.066
SRS vs Fractionated radiotherapy	0.63 (0.07-2.99)	0.630
EQD2	1.01 (0.85-1.07)	0.837
log PTV Volume	2.41 (1.12-5.91)	0.036	2.60 (1.08-6.28)	0.033

Bolded for statistically significant (p<0.05)

TSA stands for transsphenoidal surgical approach. GTR stands for gross total resection and STR stands for subtotal resection. EQD2 is the equivalent dose for a radiation treatment if it were given in 2 Gy per fraction, used to normalize for different fractionation regimens

Figure 2.

Hormonal control after radiotherapy for secretory pituitary adenomas

Toxicity: contributor to hormonal deficits but otherwise safe

Prior to radiation, 58.6% of patients had at least one pituitary hormone deficiency. Following radiotherapy, pituitary dysfunction increased with a greater percentage of patients requiring replacement of cortisol (28.6% to 58.6%), thyroid hormone (47.9% to 63.6%) and gonadal hormones (37.9% to 42.1%). The number of patients with no pituitary hormonal deficiencies went from 59 patients (42.1%) prior to radiation to 42 patients (30.0%) at last follow-up. By subgroup, 22 patients (47.8%) with functional PAs had no pituitary hormonal deficiencies prior to radiation, while only 14 patients (30.4%) had no pituitary hormonal deficiencies at last follow-up. In comparison, 37 patients (39.4%) with non-functional PAs had no pituitary hormonal deficiencies prior to radiation compared to 27 patients (28.7%) post-radiation. Changes in hormonal replacement are summarized in Table 3. On MVA, increasing need for hormonal replacement was associated with increasing number of surgeries (beta 0.41, 95% CI 0.13-0.69, p<0.01) after accounting for functional status, sex, pituitary apoplexy and KPS. Increased need for hormonal replacement post-radiotherapy was not associated on MVA with secretory status of tumor (OR 1.09, 95% CI 0.43-2.72, p=0.859), size of PTV (OR 0.50 for >= 10cc, 95% CI 0.22-1.14, p=0.098), or age at diagnosis (OR 0.97, 95% CI 0.95-1.00, p=0.071).

Table 3.

Hormonal replacement before and after radiotherapy.

Hormonal Axis Impaired	Prior to Radiation	After Radiation
Any pituitary hormonal deficit	82 (58.6%)	98 (70.0%)
Cortisol	40 (28.6%)	82 (58.6%)
Thyroid	67 (47.9%)	89 (63.6%)
Gonadal	53 (37.9%)	59 (42.1%)
Vasopressin	10 (7.1%)	10 (7.1%)

Additional toxicities were minimal including transient diabetes insipidus in 16 patients post-operatively and three patients reported unilateral muffled hearing for 2-5 months. No patients developed radiation necrosis or optic neuropathy in follow-up.

Discussion:

We report on a large single institutional experience using modern LINAC-based radiotherapy for patients with PAs. Radiotherapy is established as an adjuvant treatment for residual or recurrent PAs when surgery and/or medications are unable to provide sufficient control. Most prior literature on radiotherapy outcomes for PAs details GKRS treatment, which is not accessible to all patients and not suitable for patients that require fractionated radiotherapy. We showed that for 140 patients with functional (46) and non-functional (94) PAs, LINAC-based radiotherapy provides excellent 5-year actuarial local control and comparable long term hormonal control off suppressive medications with similar toxicities to GKRS and other prior reports.^7-9,17,18 Contrary to prior reports on increased rates of volumetric progression with functional PAs,^10,12,16 we showed no differences in local control amongst our cohort.

Prior studies have shown that local control with radiotherapy for non-functioning PAs is high,^19,20 though some reports have suggested that local control was inferior with secretory tumors.¹² Our experience shows that radiotherapy effectively provides excellent local control when complete surgical resection is not possible, even for secretory tumors. Compared to the University of Florida experience, almost all the patients in this study had definitive intent surgery, highly conformal radiation with either fixed gantry IMRT, DCA, or VMAT, and at least equivalent dose to 50.4 Gy in 1.8 Gy per fraction. In contrast, they reported only 38% of patients had curative intent surgery, 0 patients had VMAT, 55% of patients were treated with 2-3 fixed beams, and 78% had 45 Gy in 1.8 Gy per fraction (i.e. most patients had less intensive and less conformal radiotherapy).¹² Given that 49.6% of the patients in this study were treated with DCA or VMAT and none of the remaining patients were treated from fewer than 6 angles, this cohort more accurately represents the conformality of modern radiotherapy. Of note, many cases referred to radiation oncology are due to residual tumor in the cavernous sinus or adherent to optic structures limiting a complete or R0 resection or additional dose escalation for functional tumors. Furthermore, our results with LINAC-based radiotherapy using both SRS and fractionated radiation are comparable to local control rates achieved with GammaKnife.^7,9,10,17-19 Prior reports have shown 10-year local control for non-functional adenomas at 83% following SRS¹⁹ that was similar to our finding of 81% 10-year PFS with either SRS or modern, fractionated radiotherapy. Similarly, most reports of GammaKnife SRS for secretory adenomas suggest long term endocrine remission in about 51%-57% for acromegaly,^21,22 Cushing’s,^22,23 or Prolactinoma²⁴ and 29% for Nelson Syndrome.²⁵ Our findings suggest that LINAC SRS or modern, fractionated radiotherapy can yield long-term endocrine remission in 50% at 5-years and even higher rates of hormonal control in subsequent years.

It remains unclear the exact treatment and pathologic features associated with long-term tumor control. Proliferative indices (e.g., Mib-1) are an intuitive biomarker that may prove to be prognostic for local control and possibly predictive of response to SRS or dose-escalated radiotherapy. However, while one of the three local progressions was in a patient with MIB-1 of 10%, our data do not show a significant association possibly due to dearth of reporting of this feature (65% of our patients did not have MIB-1 reported on their pathology). While early post-operative SRS appears to improve outcomes for secretory adenomas,²³ there is some retrospective data to suggest that non-operative management can provide comparable outcomes for at least patients with acromegaly.²⁶ While late PFS appears slightly lower for patients treated with fractionated radiotherapy vs SRS, particularly with non-functional adenomas, this is driven by one late local failure event in the face of lower numbers that have follow-up beyond 10 years. Other factors that potentially confound differences in outcomes include the fact that many patients were ineligible for SRS due to unresectable disease, larger tumors, or closer proximity to the optic apparatus resulting in delivery of lower marginal radiation doses. We look forward to future data about hypofractionated SRS for patients that would not be eligible for single fraction SRS.

The primary chronic toxicity associated with surgery or radiation around the pituitary gland is damage to the secretory function of the normal gland requiring lifetime hormonal replacement. Even low doses of radiation have been associated with risk of hypothyroidism following head and neck radiation,²⁷ while central nervous system radiation is a known cause of hypopituitarism.^28-30 Estimates range on rates of hypopituitarism though long term follow-up shows that the majority of patients with PAs will develop some if not multiple hormonal deficiencies. Prior reports with GammaKnife SRS showed 23-31% of patients without pre-treatment hypopituitarism will develop new deficiencies.^31-33 Similarly, 28% of patients in our cohort with normal pituitary function developed hypopituitarism after LINAC radiotherapy. However, many of these deficits are present before radiation as a result of damage from the tumor or surgical resection. Our results are in line with prior work though also highlight that prior to radiation 58.6% of patients had at least one pituitary hormonal axis disrupted suggesting that the cause of long-term pituitary hormone deficiencies is multifactorial and can be independent of radiotherapy utilization. Long term follow-up for these patients is essential as our cohort and prior work has shown that hypopituitarism is a common side effect from PAs and their operative and non-operative treatments. Further work is needed to elucidate the mechanisms of radiation-induced hypopituitarism now that modern radiotherapy can significantly spare adjacent structures.

This study has multiple limitations. The retrospective nature of this study limits comparisons between different types of treatment received from surgeries to method of radiotherapy. Few progression events limited analysis of associations with patient, tumor, or treatment characteristics. Furthermore the number of patients in sub-groups (e.g. functional PAs), may have increased the risk of type II error. Furthermore, an increasingly common treatment that has been adopted both at our institution and others recently is hypofractionated radiosurgery. However we limited our study to patients with at least 3 years of MRI follow-up, which excluded patients treated at our institution in the last 2 years with 25 Gy in 5 fractions over one to two weeks with frameless, LINAC-based SRS.

Conclusions:

LINAC-based radiotherapy for patients with PAs was safe and effective for hormonal control and local tumor control with comparable results to prior studies using GKRS, a treatment machine less ubiquitous than LINACs. Unlike previous reports, no difference in local tumor control was noted for secretory versus non-secretory tumors, possibly due to refined radiation technique and higher doses used for patients in this study. Further work is needed to elucidate factors to increase long term hormonal control in secretory tumors and to decrease the high rates of hormonal deficiencies following single and combined modality management of PAs.

Abbreviations:

ACTH:

Adrenocorticotropic hormone

ANOVA:

Analysis of Variance

CI:

Confidence interval

DCA:

Dynamic conformal arcs

DFS:

Disease Free Survival

EQD2:

Equivalent dose in 2 Gy per fraction

GH:

Growth hormone

GKRS:

GammaKnife radiosurgery

HR:

Hazard ratio

IMRT:

Intensity modulated radiation therapy

IQR:

Inter-quartile range

KPS:

Karnofsky performance status

LC:

Local control

LINAC:

Linear accelerator

mFU:

Median follow-up

MRI:

Magnetic resonance imaging

MVA:

Multi-variable analysis

OR:

Odds ratio

PAs:

Pituitary adenomas

PTV:

Planning target volume

SRS:

Stereotactic radiosurgery

TSH:

Thyroid-stimulating hormone

UVA:

Univariate analysis

VMAT:

Volumetric modulated arc therapy