Secondary malignancy following stereotactic radiosurgery for benign neurologic disease: A cohort study and review of the literature

Abstract

Radiation-associated malignancy and malignant transformation are risks associated with stereotactic radiosurgery (SRS); however, incidence is uncertain. The purpose of our study is to present the rate of radiation-associated malignancy and malignant transformation at our institution. After IRB approval, we undertook a retrospective cohort study evaluating patients treated with Gamma Knife® or linac-based SRS between 1990 and 2014 for benign CNS pathology with at least 5 years of clinical follow-up. Risk of transformation was calculated using the Kaplan-Meier method. A total of 273 patients met selection criteria. Median clinical follow-up after SRS was 11 years (range 5-27). Over 3,216 patient-years following SRS, we found zero cases of radiation-associated malignancy and two cases of radiation-associated malignant transformation for a crude rate of 0.73% or case rate of 0.62 per 1,000 patient-years. The Kaplan-Meier risk of malignant transformation at 5, 10, and 15 years was 0.4% (95% CI 0.05% 2.6%). These data support the continued use of SRS for benign intracranial pathology without significant concern for secondary malignancy.

Introduction

Stereotactic radiosurgery (SRS) is a highly conformal image-guided delivery of ionizing radiation in one to five fractions for the treatment of malignant and benign neurologic disease [1]. Due to its efficacy and minimal invasiveness, the practice of SRS has dramatically expanded in recent years [2,3]. For example, Medicare billing data suggest a nearly ten-fold increase in utilization of SRS with over 15,000 codes billed in 2011 [4]. In 2017, the Leksell Gamma Knife Society reported that over 1.2 million people had been treated with Gamma Knife SRS [5]. It is unknown how many patients have been treated with linac-based SRS. Interestingly, some authors estimate that at least 80,000 patients have been treated with SRS for benign disease [6,7]. With the advent of SRS in the treatment of benign tumors and functional conditions, including treatment-refractory epilepsy, tremor, trigeminal neuralgia, and psychiatric disease, special concerns exist regarding the risk of radiation-associated malignancy in the treatment of patients without cancer [8–12].

In the last two decades, several reports have been published describing cases of malignant transformation of benign tumors following SRS [13–16]. Patel and Chiang recently reviewed the literature and described a total of 36 cases of SRS-induced neoplasms [17]. These authors extrapolated that the risk of SRS-induced neoplasm was 0.04% at 15 years based on an estimation of the total worldwide number of patients treated with SRS for a benign disease [17]. Although this risk is exceedingly small, the possibility of contributing to the development of a side effect of such seriousness as a malignancy warrants careful consideration. The authors’ approach of using SRS prevalence as the denominator for risk estimation has significant methodological limitations, and thus the risk of developing new neoplasms after SRS remains poorly understood [18]. Therefore, we sought to describe the risk of radiation-associated malignancy as well as malignant transformation after SRS for benign lesions at 5, 10, and 15 years.

Materials and Methods

After institutional review board (IRB) approval, we undertook a retrospective cohort study evaluating patients treated with SRS between 1990 and 2014 for benign central nervous system pathology. Informed consent was waived by the IRB due to minimal risk. All patients were followed clinically for a minimum of 5 years for inclusion in this study. Patients who expired from other causes prior to 5 years of follow-up were excluded. Patients with progression of their benign condition were retained in analysis, so long as they achieved at least 5 years of follow-up. For each patient, the electronic medical record was assessed for demographic and clinical details. Radiation-associated malignancy was defined based on the Cahan criteria as a new histologically-distinct primary CNS malignancy within the 2 Gy isodose line arising in a patient without genetic predisposition to tumor development at least 4 years after initial radiation [19]. Malignant transformation was defined as the dedifferentiation of a benign tumor to a tumor with malignant features of the same histologic cell of origin. The primary study outcome was the five-year, ten-year, and fifteen-year incidence of radiation-associated malignancy and malignant transformation after SRS.

For the review of the literature, we queried PubMed in November 2019 for the terms “Secondary malignancy” OR “Malignant transformation” AND “Stereotactic radiosurgery” OR “SRS” OR “Radiosurgery” OR “FSRS”. The field “Title/Abstract” was utilized. Fifty-three articles were returned by the query. Both retrospective and prospective articles containing original patient data reporting risk of secondary malignancy and malignant transformation after SRS, as well as meta-analyses, were eligible for inclusion. Review articles, editorials, case reports, and opinion pieces were excluded but evaluated for references of eligible studies. Studies were also excluded if median follow-up was less than 5 years. After application of inclusion and exclusion criteria, 8 studies were reviewed for sample size, follow-up, and crude rate of secondary malignancy and malignant transformation.

Baseline characteristics and treatment details were summarized with descriptive statistics. The Shapiro-Wilk test was used to assess for normality of continuous variables. Covariates that lacked a normal distribution were described by median and range. Time to event outcomes were analyzed by Cox proportional hazards regression models. Proportionality was assessed for all covariates by the Kolmogorov-type supremum test with 1000 simulations and by graphing martingale score residuals versus time. The rate of malignant transformation was determined by the Kaplan-Meier method. Multivariable analysis was not performed due to the small number of events to avoid overfitting. Malignancy-free survival was characterized by fitting Kaplan-Meier curves. All confidence intervals (CI) were reported at 95% and all p values were reported as two-sided, with significance defined at a level of p<0.05. All statistical analysis was performed using SAS version 9.4 (Cary, NC); plots were generated using GraphPad Prism version 7 (La Jolla, CA).

Results

A total of 635 consecutive patients treated with radiosurgery for benign disease were evaluated for study enrollment. After excluding 362 patients due to insufficient follow-up, 273 patients met selection criteria. Patient demographics and characteristics are displayed in Table 1. Treated diagnoses included arteriovenous malformation (AVM), pituitary adenoma, World Health Organization grade I-II meningiomas, vestibular schwannoma (VS), paraganglioma, angiofibroma, chordoma, craniopharyngioma, epilepsy, ganglioglioma, hemangioblastoma, hemangiopericytoma, oligodendroglioma, papilloma, parotid adenoma, tremor, and trigeminal neuralgia. The median age at the time of treatment was 44 years (range 2-82); both pediatric (n=33) and adult (n=240) patients were included. Median dose was 18 Gy (range 6-140 Gy), and median number of fractions was 1 (range 1-5). Median clinical follow-up after treatment was 10.5 years (range 4.7-27.0). At 15 years, 40 patients (54%) continued to maintain radiologic follow-up.

Table 1.

Patient demographics, indications for radiosurgery, and radiosurgical characteristics.

Characteristic	Median/incidence	Range/frequency
Age (yrs)	44	2-82
Sex (female)	172	60%
Race
White	242	90%
Black	27	10%
Dose (Gy)	18	6-140
Fraction	1	1-5
Diagnosis	Number of Cases	Proportion (%)
AVM	100	37%
Pituitary adenoma	67	25%
Meningioma	39	14%
Vestibular schwannoma	32	12%
Paraganglioma	13	5%
Angiofibroma	1	0.40%
Chordoma	2	0.70%
Craniopharyngioma	1	0.40%
Epilepsy	1	0.40%
Ganglioglioma	2	0.70%
Hemangioblastoma	2	0.70%
Hemangiopericytoma	1	0.40%
Oligodendroglioma	1	0.40%
Papilloma	3	1.00%
Parotid Adenoma	1	0.40%
Tremor	1	0.40%
Trigeminal Neuralgia	6	2.00%

Abbreviations: Gray (Gy), arteriovenous malformation (AVM)

Over 3,216 patient-years following SRS, no cases of radiation-associated malignancy were observed. Two cases of radiation-associated malignant transformation were recorded. For the overall follow-up period, the crude incidence of radiation-associated malignancy and malignant transformation was 0.73%, and the case rate was 0.62 cases per 1,000 patient-years. The Kaplan-Meier calculated risk of malignant transformation and radiation-associated malignancy was 0.4% at 5 years (95% CI 0.05% 2.6%), with 271 patients at risk; 0.4% at 10 years (95% CI 0.05% 2.6%), with 141 patients at risk; and 0.4% at 15 years (95% CI 0.05% 2.6%), with 74 patients at risk (Figure 1). This risk increased at 20 years to 1.8% (95% CI 0.4%-8.9%), with 34 patients remaining at risk in our cohort. Considering just the sporadic VS cohort (n=35), the crude risk of malignant transformation was 2.9%. On single predictor Cox proportional hazards regression, clinicopathologic factors, including age, sex, race, ethnicity, diagnosis, dose, and fractionation, were either not significantly associated with malignant transformation or did not satisfy proportional hazards assumptions.

Figure 1.

Kaplan-Meir survival plot of radiation-associated malignancy and malignant transformation free survival following stereotactic radiosurgery for benign neurologic disease. Number of subjects at risk is listed on the x-axis.

Case 1

The first patient who developed malignant transformation was a 28-year-old Hispanic female originally presenting with throbbing radiating headache associated with nausea and progressive left-sided hearing loss. Physical exam was notable for left-sided numbness in the V1, V2, and V3 distributions, moderate sensorineural hearing loss, bilateral foot numbness, ataxia, and diplopia. Head CT from an outside facility demonstrated a 3 x 2.8 x 4.5 cm diffusely enhancing mass arising from the left tentorial edge at the petrous apex with brainstem compression, marked midline shift, and obstructive ventriculomegaly. MRI revealed an approximately 6 cm left posterior fossa mass extending into the internal auditory canal. The patient underwent staged translabryinthine resection two weeks apart; histologic diagnosis by Vanderbilt pathology was benign schwannoma. At the time of her second staged resection, a small amount of tumor involving the cerebellar peduncle and tumor involving the facial nerve was not removed due to an intraoperative determination of unsafe resection. MRI obtained 3 months later was concerning for interval progression with the tumor now measuring 1.8 x 1.4 x 1.5 cm. The patient was then treated with linac-based SRS delivered in 3 fractions to a total dose of 21 Gy. Serial imaging over 13 years showed no evidence of recurrence. Fifteen years after SRS, she presented with increasing left facial numbness and acute headaches. CT and MRI demonstrated a left cerebellopontine angle mass measuring 3.6 x 2.2 cm intimately involving the midbrain and pons with marked increase in size compared to imaging two years prior. She underwent maximal safe translabyrinthine resection complicated by recurrent cerebrospinal fluid leak requiring surgical repair. Final pathology demonstrated neoplastic proliferation of markedly atypical spindle cells with high Ki-67 index and frequent mitotic figures (up to 14 per high powered field) along with pseudopalisading necrosis. BCL-2, CD99, CAM 5.2, SMA, and reticulin stains were positive, consistent with a malignant peripheral nerve sheath high grade sarcoma and considered by our pathologists to be radiation-induced transformation. MRI after surgery revealed residual tumor at the superior extent of the surgical bed associated with the anterior free edge of the tentorium and marked improvement in vasogenic edema and mass effect. Her post-operative management was complicated by septicemia and meningitis, protracted rehabilitation with diminishing performance status, and significant dysphagia requiring percutaneous endoscopic gastrostomy. Given her clinical deterioration and goals of care, she was ultimately discharged to hospice.

Case 2

The second case of malignant transformation was a 68-year-old Caucasian male who presented with chronic hearing loss. Audiogram demonstrated profound left-sided sensorineural hearing loss. MRI revealed bilateral VSs consistent with delayed-onset neurofibromatosis type II. The left internal auditory canal mass measured 1.5 x 1.2 cm with local mass effect on the pons and cerebellar peduncle, and the right cerebellopontine angle mass measured 0.7 x 0.7 cm. The patient was recommended to undergo surgical resection of the left presumed VS vs. low grade neurofibroma; however, he elected for Gamma Knife® radiosurgery at an outside hospital to the left mass and observation of the right mass. Three years later, he presented with progressive imbalance and left-sided facial weakness. MRI revealed unequivocal enlargement of the treated tumor with significant brainstem compression and edema. He underwent translabyrinthine subtotal resection due to severe adherence of the tumor to the brainstem and cerebellum associated with high vascularity and edema. Pathology demonstrated a neoplastic proliferation of spindle cells with variable Ki-67 index and mitotic activity, and a diagnosis of recurrent nerve sheath tumor with elevated proliferation index was rendered. Four months later, MRI demonstrated interval progression. The patient elected for observation. Serial MRI over the next eight months demonstrated explosive enlargement and a tumor diameter of 5 cm with brainstem compression associated with neurologic decline. The patient then underwent a retrosigmoid resection and debulking; pathology showed frequent mitotic figures, moderate to high Ki-67 rate, and positive H3K27me3 staining consistent with a malignant peripheral nerve sheath sarcoma. Throughout this period, the untreated right-sided schwannoma demonstrated indolent mild to moderate growth. The patient continued to decline post-operatively with high flow gastrointestinal bleed with cardiac instability, and expired on post-operative day 3.

Discussion

In this study, we demonstrate that the five-year, ten-year, and fifteen-year risk of secondary malignancy and malignant transformation following SRS for benign neurologic disease is reassuringly low at 0.4% (95% CI 0.05% 2.6%). Over 27 years of institutional follow-up after SRS encapsulating 3,216 patient-years, we identified no cases of radiation-associated malignancy and only two cases of malignant transformation for a crude risk of 0.74%, resulting in a rate of only 0.62 cases per 1,000 patient-years. These data add to the growing literature on the safety of SRS in the treatment of non-malignant disease, and provide additional evidence to the physician’s armamentarium for counseling patients on the risks of secondary neoplasm after SRS, including an increasing risk over time. This risk may be related in part to the volume of normal tissue irradiated and, because SRS is a focal technique, the risk of secondary neoplasm may be attenuated compared to other methods of external beam radiotherapy utilizing larger fields [20,21]. For instance, Minniti et al. reported a 2.0% risk of secondary malignancy at 10 years following conventional three-field radiotherapy for pituitary adenoma, which is significantly higher than our rate [22].

Other institutions have also assessed second/transformed malignancy risk, and reported a crude rate of secondary malignancy and malignant transformation after SRS between 0% and 0.7% (Table 2). Pollock and colleagues evaluated this risk in patients treated with single-fraction SRS at the Mayo clinic [23]. In their study, 8 cases of malignant transformation out of 909 patients were recorded for a 0.9% risk, consistent with our findings. Interestingly, the authors suggested that meningioma pathology may be associated with increased risk of transformation (HR 11.72, 95% CI 1.44-96.15, p=0.02) [23]. No radiation-induced tumors were observed in over 13,000 patients treated with SRS at the University of Pittsburgh, although three radiation-induced tumors were detected in 1309 patients treated at the University of Virginia [24,25]. Recently, an international pooled analysis of 5000 patients treated with SRS for benign disease calculated a 0.0006% risk of malignant transformation and 0.0002% risk of radiation-induced malignancy [26]. Due to the small number of cases in the literature, few clinical parameters, aside from genomic instability, have been identified as risk factors for secondary malignancy or malignant transformation. From these data and our findings, the risk of malignant transformation and malignancy following SRS should not be over-emphasized in determining each patient’s risk:benefit calculus, and clinical decision-making for patients with non-malignant disease contemplating SRS should include the non-zero, but minimal, risk estimate for these rare unfortunate events.

Table 2.

Literature review of the risk of secondary malignancy or malignant transformation following stereotactic radiosurgery.

Reference	Histology	n	Median Follow-up (years)	Crude Rate of Secondary Malignancy and Malignant Transformation	Latency (years)
Pollock et al., 2017 [23]	Variousa	1142	9.0	0.7% (8 cases)	2.8-13.8
Wolf et al., 2019 [26]	Variousb	4905	8.1	0.06% (3 cases)	8.7-12.8
Hasegawa et al., 2013 [43]	VS	440	12.5	0.3% (1 case)	5.5
Rowe et al., 2007 [30]	Variousc	4877	5.2	0.02% (1 case)	8.0
Frischer et al., 2018 [44]	VS	426	5.1	0.2% (1 case)	8.6
Birckhead et al., 2016 [45]	NF2-associated meningioma	15	9.3	0% (0 cases)	NA
Gao et al., 2019 [46]	NF2-associated meningioma	35	8.0	0% (0 cases)	NA
Starke et al., 2014 [25]	AVM	1309	7.8	0.2% (3 cases)	8-19

^aArteriovenous malformation, meningioma, vestibular schwannoma, pituitary adenoma, glomus tumor

^bVestibular schwannoma, meningioma, arteriovenous malformation, trigeminal neuralgia, pituitary adenoma, hemangioblastoma, other schwannoma

^cArteriovenous malformation, meningioma, vestibular schwannoma, cerebral metastasis, other tumor, other pathology

Abbreviations: vestibular schwannoma (VS), neurofibromatosis type 2 (NF2), arteriovenous malformation (AVM), non applicable (NA)

Regarding sporadic VS, we report a crude risk of malignant transformation of 2.8%. In a 2015 review, 8 cases of malignant transformation of VS following SRS were compared to 18 cases of primary malignant VS [27]. Due to the similar frequencies, these authors argued that malignant transformation may arise from the natural progression of VS even after SRS, rather than causally from SRS [27]. Other groups have hypothesized that the risk of malignant transformation after SRS may be similar to de novo transformation risk [26,28–30]. In Case 1, the latency period of 15 years is significantly greater than the largest latency period in the cases reviewed by Maducdoc and colleagues, suggesting a lower likelihood of a causal relationship with radiation [27]. Furthermore, there are at present no studies that have demonstrated a definitive causal relationship between radiosurgery and malignant transformation of VS, and there are no published guidelines that define malignant transformation, increasing the difficulty of diagnosis [27,31,32].

It is also notable in Case 1 that two surgeries preceded SRS. Data from the Mayo Clinic suggest there may be an increased risk of malignant transformation after resection followed by SRS versus SRS alone (HR 14.56, 95% CI 1.79-118.33, p=0.01) [23]. Four cases of malignant transformation of VS have been reported following resection alone [27]. Various hypotheses of this phenomenon have been proposed, including electrocautery, post-surgical inflammation and edema within the tumor microenvironment, and natural progression [27,32,33]. There is also some evidence on the role of long-term aspirin and reduced VS growth [34]. Further study of this phenomenon and potential strategies, such as anti-inflammatory medications, is needed.

The risk of malignant transformation may be increased in the setting of heritable genomic instability, as suggested by Case 2. The association between malignant nerve sheath tumor and neurofibromatosis type II, hypothesized based on Knudson’s two-hit hypothesis, is controversial [35–39]. Two case series have cast doubt on this association, although a study of population-level data suggests that an increased risk exists. [37,40,41]. Determining causality between genetic predisposition and radiosurgery is challenging, given the natural history of neurofibromatosis type II and background incidence of malignant tumors [42]. Nonetheless, in the case presented here, the stability of the contralateral untreated tumor and change in disease biology after radiosurgery suggest a higher likelihood of a causal role. Caution may be warranted in this population, although further investigation is needed.

Limitations of this study include the retrospective nature and small sample size. Although median clinical follow-up of our cohort was over ten years, secondary malignancy and malignant transformation may occur outside the follow-up period of the present study, with latency reported in the literature up to 19 years (Table 2). Thus, our study may underestimate late risks. Additionally, while patients maintained long-term clinical follow-up, minimum radiologic follow-up was not mandated for inclusion in our study, although the majority of patients continued to maintain radiologic follow-up beyond 15 years. Furthermore, 57% of our potential cohort was excluded due to insufficient follow-up, raising concerns for selection bias. Thus, the incidence of patients with secondary neoplasm may be underreported here, especially if there was an association between follow-up and our outcomes of interest. Finally, the number of events in our cohort was too small for powered statistical analysis of predictive clinicopathologic factors. Thus, while our study adds to the growing literature evidencing the low rate of secondary neoplasm following SRS, our report does not offer data that provide opportunities to reduce this minimal risk further.

Conclusions

We report a low rate of radiation-associated malignancy and malignant transformation following SRS for benign neurologic disease after 27 years and 3,216 patient-years of institutional experience. This risk may be elevated in patients with genetic predisposition to tumorigenesis. Physicians counseling patients on the risk of secondary neoplasm following SRS for benign disease should disclose the non-zero, but minimal, risk associated with radiosurgery as evidenced by these findings and by the growing body of evidence. Although SRS is an attractive option for patients with neurofibromatosis type II due to the minimally invasive approach without operative risk, physicians should counsel patients on the potentially elevated risk of secondary neoplasm in this population compared to sporadic VS. Future investigations should continue to quantify this risk in patients with over thirty years of follow-up following radiosurgery, but this risk should not discourage the use of SRS for benign conditions.