Vision loss in glioblastoma: Disease mimicking presumed therapeutic toxicity

Abstract

Glioblastoma is the most common and lethal form of primary brain cancer. In the recurrent setting, bevacizumab is a common choice for salvage therapy. Loss of vision in patients initially treated with radiation at the time of diagnosis and later treated with bevacizumab at time of recurrence has been reported, and presumed to be a treatment-related optic neuropathy. Strikingly, only 1 case report described a postmortem biopsy to rule out tumor involvement of the optic tracts. We report the first case of recurrent glioblastoma infiltrating the prechiasmatic and chiasmatic optic nerve, which at the time of vision loss was presumed to be secondary to bevacizumab. It is noteworthy that the MRI findings in the previously reported bevacizumab/radiation-induced optic neuropathy cases (without pathology follow-up) are comparable to our patient.

It is well recognized that delayed radiation-induced neurotoxicity can occur if cranial nerves are included in the radiation treatment field, with risk being dose-related.^1,2 Relative to this, in vitro and in vivo studies have shown that bevacizumab can act as a radiation sensitizer in combined, as well as sequential, treatment protocols.³

After the introduction of bevacizumab as a glioblastoma treatment modality, Kelly et al reported 3 glioblastoma patients who received upfront radiation and chemotherapy and later experienced vision loss following salvage therapy with bevacizumab.⁴ These authors postulated that the vision loss was caused by the inhibition of VEGF-dependent repair of radiation-induced sublethal damage of normal tissue.⁴ The radiation treatment fields of these patients involved the optic nerves and chiasm, leading to a presumption of bevacizumab/radiation-induced optic neuropathy. Enhancement signal of the optic nerves on postcontrast T1-weighted MRI was reported for 2 of the 3 cases. However, postmortem biopsy was not done to exclude possible tumor involvement in the optic nerves and chiasm. Sherman et al reported 6 cases with radiation fields encompassing the optic chiasm and nerves that incurred vision changes after bevacizumab initiation.⁵ Enhancement on postcontrast T1-weighted MRI was detected on the optic nerve for 2 of 6 cases (1 bilateral and 1 left optic nerve). An autopsy was only done in 1 case, with negative tumor involvement in the optic nerve, leading to the conclusion of a therapy-induced optic neuropathy. Most recently, 2 additional cases were reported by Fu et al⁶_; both cases lacked postmortem evaluation.

Below we present the first reported case of recurrent glioblastoma infiltrating the optic nerve. The clinical course of our patient, as well as the MRI findings, were almost identical to those found with the presumed neurotoxicity in the aforementioned bevacizumab case reports.

Case Report

In September 2014, a 56-year-old woman presented with seizures. An MRI demonstrated a large area of T2-weighted hyperintense signal throughout the medial right temporal lobe. Gross total resection was accomplished in October 2014. Subsequent pathology confirmed grade 3 anaplastic astrocytoma with wild-type isocitrate dehydrogenase 1 and 2 (Sanger sequencing), unobtainable MGMT, and intact 1p/19q.

Standard radiochemotherapy was completed in December 2014, followed by 1 cycle of adjuvant temozolomide. Radiation was given at a total dose of 59.4 Gy (28 fractions of 1.8 Gy/fraction). The right optic nerve received an average of 44.2 Gy +/- 7.8 Gy; one third of the optic nerve received approximately 51 Gy. An MRI in February 2015 demonstrated a markedly increased area of enhancement in the right posterior medial temporal lobe with extensive surrounding T2 hyperintensity. A near gross total resection was accomplished, and pathology was upgraded to glioblastoma with unmethylated MGMT. The patient subsequently received 2 months of tumor treatment field therapy (Optune^®) before another progression of disease in June 2015. Bevacizumab was then initiated at the dose of 10mg/kg every 2 weeks. After two months of therapy, the patient developed vision loss from her right eye. An MRI at the time showed mild enlargement of the right prechiasmatic optic nerve with associated T2 hyperintensity (Fig. 1A) and trace T1 enhancement. This was thought to be bevacizumab-induced optic neuropathy based on the existing literature,^4,5 resulting in discontinuation of bevacizumab. Ophthalmological exam revealed no light perception on the right; dilated fundoscopy demonstrated an atrophic right nerve. Soon after, the patient’s condition deteriorated rapidly despite retreatment with radiation. Her final MRI showed significant tumor involvement in the right hemisphere with extension across the splenium of the corpus callosum. Increased T2/fluid-attenuated inversion recovery (FLAIR) hyperintensity was noted in the right prechiasmiatc and chiasmatic optic nerve (Fig. 1B) with associated T1 nodular enhancement. Postmortem autopsy (April 19, 2016) showed glioblastoma involvement in the right optic nerve (Fig. 2) and the right lateral geniculate nucleus (not shown).

Fig. 1.

Compound oblique sagittal T2/FLAIR (fluid-attenuated inversion recovery) reconstruction images through the right optic nerve: A) September 2015: Increased signal and mild swelling in the prechiasmtic right optic nerve. B) February 2016: Intraorbital increased signal in the right optic nerve now tracking into the optic chiasm (arrows). Also seen is increased T2 signal extending across the callosal splenium extending into the parietal and occipital lobes consistent with disease progression.

Fig. 2.

Histology of the optic nerve from the patient’s postmortem biopsy. A) Whole mount section of optic chiasm stained with haemotoxylin & eosin (H&E) (upper panel A) and luxol fast blue with periodic acid-Shiff (LFB/PAS) (lower panel A). To the right there is loss of myelin staining secondary to tumor infiltration and necrosis. (Note: The line running through panel A in the H&E (upper) as well as the LFB/PAS (lower) represent a necessary cut in the photomicrograph order to encompass the entire nerve at low (2x) magnification; cuts were purposely made in different places. B) H&E stained section of the right side of the optic chiasm at original magnification of 20x. Infiltrating neoplastic astrocytes (arrows) and necrosis (*) have replaced myelinated axons of the optic nerve.

Discussion

Although infiltration of the optic nerve by recurrent glioblastoma represents an expected natural history for this disease, our report is the first documented case of which we are aware. Relative to this, neurotoxicity of the optic nerve from bevacizumab and/or radiation treatment has been reported in 11 cases; however, only 1 included postmortem findings to evaluate the possibility of tumor involvement.^4,5 Interestingly, of 11 total reported cases, 10 involved either the frontal or temporal lobes, raising the anatomic possibility of direct spread to the optic nerve.^4–6

Our patient presented with right frontal transformed glioblastoma and developed vision loss after 2 cycles of bevacizumab in the recurrent setting. Enhancement at the right optic nerve was detected on postcontrast T1-weighted MRI. As the initial radiation field involved the right optic nerve and the patient was receiving bevacizumab at the time of vision loss, bevacizumab-induced optic neuropathy was presumed, and treatment was discontinued. However, postmortem examination confirmed glioblastoma of the right optic nerve. In retrospect, upon reviewing the FLAIR MRI in February 2016, the tumor migrated dorsally to the right lateral geniculate nucleus, and tracked anteriorly along the optic tract towards the optic chiasm and into the prechiasmatic right optic nerve. In September 2015, the migration of the tumor was first detected and paralleled her vision loss; however, these MRI findings were not present a month prior. As is evident, the ability to distinguish the difference between tumor infiltration from bevacizumab-induced optic neuropathy is challenging. Indeed, we are not aware of any radiological procedures to distinguish between therapy-induced neuropathy and tumor invasion.

Conclusion

Our case demonstrates that optic involvement of recurrent glioblastoma might mimic radiation-induced and/or bevacizumab-induced optic neuropathy, resulting in the premature discontinuation of a therapeutic modality. Thus, radiological changes and clinical symptoms should be integrated into an anatomical consideration of the potential for disease invasion.

Funding

Sources of Support: This work was supported by the National Institutes of Health National Cancer Institute CORE grant (grant number 4P 30 CA014520).