*3.3. Frequent Adrenal Insu*ffi*ciency in Patients with Corticosteroid-Refractory Cerebral Radionecrosis*

Treatment with dexamethasone was started at a median of four months (range 0–64) after radiation therapy and continued for a median of 141 days (range 18–699). The median maximum dose of dexamethasone was 8 mg (range 4–100), which was reduced stepwise. Dexamethasone was stopped at a median of 61 days after the start of bevacizumab treatment (range −4–372). In 10 patients, the tapering of dexamethasone consisted of a dose reduction to 0.5–1 mg and subsequent switching to hydrocortisone before the analysis of cortisol. In the seven other patients, dexamethasone doses were reduced to either 0.25 mg daily or 0.5 mg every other day, without switching to hydrocortisone. Cortisol analysis revealed an adrenal suppression (<7.25 μg/dL) in five patients (Figure S1). One of these patients presented only a slight reduction of serum cortisol (patient number 5: cortisol 5.7 μg/dL); the other four patients had a marked pathological cortisol value. Additional confirmation of adrenal function via ACTH stimulation was available in 11 patients (two patients with AI and nine patients without AI, as indicated by basal cortisol values) [21]. Notably, the results of all ACTH stimulation tests matched the results of the basal cortisol analysis. Analysis via *t*-test did not show a significant difference in age, duration of dexamethasone treatment, or time from radiation therapy to start of dexamethasone or to start of bevacizumab between the group of patients with normal adrenal function and the group of patients with AI. The Levene test of variation showed an imbalance of variation for age at cortisol analysis and time from radiation therapy to initiation of dexamethasone/bevacizumab, which could have confounded the results. A comparison of the two groups is given in Table 1. Regression analysis did not show a correlation between cortisol value and duration of dexamethasone treatment or patient age (Figure 2).


**Table 1.** Comparison of the patients with and without adrenal insufficiency.

Data are displayed as mean ± standard deviation. *p*-values were calculated using an independent samples *t*-test.

**Figure 2.** Correlation of cortisol levels with clinical data. Regression analysis of (**A**) patients' serum cortisol level and their age and (**B**) patients' serum cortisol level and duration of dexamethasone treatment did not reveal a significant correlation. The red line indicates the lower limit of the normal value of 7.25 μg/dL (200 nmol/L) cortisol. The linear regression line is shown in black.

#### **4. Discussion**

We here report that AI is a frequent condition in patients undergoing bevacizumab treatment for cerebral radiation necrosis, which was detectable in approximately 30% of patients after the cessation of dexamethasone. Our study highlights the need for cortisol testing in brain tumour patients, independently of the treatment duration of corticosteroids.

The frequency of AI in our cohort is in line with a study of patients receiving dexamethasone as a supportive drug for high emetogenic chemotherapy, which revealed a rate of AI of 15% [22]. We further confirm the potent effect of bevacizumab in treating radiation necrosis in our cohort of 17 patients, which has previously been published in a randomized cohort of 14 patients [18]. While severe side effects of bevacizumab have been reported when therapy was administered over longer periods and as part of tumour-targeted treatment regimens with chemotherapy, in short-course and reduced-dose treatment, the side effects profile is most likely significantly less severe [13,18,23].

Almost all patients with intracranial tumours receive dexamethasone during their course of treatment. Dexamethasone is used to treat edema during surgery and radiation therapy and as add-on therapy to chemotherapy to improve the neurological deficit or increase drug compatibility. In patients with recurrent tumours, especially after a second round of radiation therapy, dexamethasone is often required at least temporarily. However, there is increasing evidence for a negative influence of corticosteroids on glioma patients [24]. Dexamethasone-induced leukocytosis is associated with shorter survival and increased risk for lymphopenia [25–27]. Therefore, the lowest possible dose of

dexamethasone should always be administered. Nevertheless, balancing dexamethasone doses with edema-associated morbidity can be challenging in daily practice. Patients treated with bevacizumab as part of a chemotherapy regimen or to treat radiation necrosis are the exception in which dexamethasone can frequently be reduced and eventually stopped. Our retrospective analysis revealed a considerable proportion with AI after prior dexamethasone treatment. This is of the utmost clinical importance because AI symptoms can easily be overlooked. Fatigue is one of the most frequent complaints of tumour patients undergoing radiation therapy, but it can also be indicative of AI. We therefore propose testing for AI when terminating dexamethasone treatment with a low threshold, especially in elderly patients and patients who have received dexamethasone over a long period. Additionally, in times of stress (surgery, infection) or unspecified clinical deterioration, cortisol testing should be included in the clinical workup. Notably, none of the patients in our cohort permanently discontinued corticosteroids without prior evidence of sufficient adrenal function. Therefore, specific clinical symptoms of AI were not detected.

The main limitation of our analysis is the small and select sample size, with potential bias for overrating the frequency of AI with regard to the general population of brain tumour patients. Cortisol analysis is not routinely included in the laboratory workup of glioma patients and was only available in 17 of the 40 initially identified patients with bevacizumab treatment for cerebral radiation necrosis (Figure S1). The small sample size may also be the cause of the rather unexpected lack of a significant correlation between the duration of dexamethasone treatment and AI. However, this result is in line with findings in other studies. AI was common in patients with corticosteroid treatment for glomerular disease, but was not predicted by daily dose or duration of treatment [28]. In patients with rheumatoid arthritis treated with corticosteroids, the duration of treatment was also not significantly associated with AI [29].

Another limitation is the bias of a retrospective analysis. It is possible that cortisol was only determined when a higher risk of AI was anticipated and, thus, the proportion of AI may be overestimated. However, cortisol testing is not a common element in routine serum analyses in nonpituitary brain tumour patients, and therefore AI may be greatly underdiagnosed. Further prospective data collection is necessary to estimate the true rate of AI in glioma patients.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2077-0383/8/10/1608/s1, Supplementary Table S1. Patient collective. Supplementary Figure S1: Consort diagram of the study population.

**Author Contributions:** Conceptualization, M.V.; formal analysis, investigation and data curation, M.V., A.B., M.W. and E.S.; writing—original draft preparation, M.V. and M.W.R.; writing—review and editing, M.V., M.W.R., J.B., M.-T.F., C.M.R., J.P.S. and J.B.; visualization, M.V. and M.W.R.

**Acknowledgments:** The Senckenberg Institute of Neurooncology is supported by the Senckenberg Foundation. M.W.R. has received fellowships from the UCT Frankfurt and funding from the Frankfurt Research Funding (FFF) Clinician Scientists Program.

**Conflicts of Interest:** J.P.S. has received honoraria for lectures or advisory board participation or consulting or travel grants from Abbvie, Roche, Boehringer, Bristol-Myers Squibb, Medac, Mundipharma, and UCB.
