Dexamethasone is frequently given to alleviate signs and symptoms related to vasogenic cerebral edema in patients with brain tumors [1-3]. While many imaging studies have demonstrated its efficacy in reducing peri-tumoral edema [5-7] and improving peri-tumoral CBF , its clinical benefits have not been systematically studied. Rather, it has only been suggested that focal neurologic impairments respond less readily to steroids than global signs or symptoms of high ICP . This report demonstrates that focal deficits respond to steroids in only about a third of cases, but that this response depends on deficit type and, in all cases, predicts long-term improvement. Secondarily, intensity and duration of steroid therapy do not clearly influence the likelihood of response.
The overall rate of improvement in neurologic deficits following steroid administration was 38.5% [2, 12-16]. Even so, the likelihood of response appears to vary widely depending on clinical condition or, in this instance, deficit type. Impairments in motor function responded most readily to steroids, while language and visual field dysfunction were relatively less likely to improve. Patients’ higher sensitivity to motor impairment than visual field loss may lead to earlier presentation and, thus, this favorable response profile. Similarly, progressive functional loss may reach a threshold beyond which improvement without surgery becomes exceedingly unlikely, again favoring earlier presentation. Even so, 82% of all patients in this study displayed symptomatic improvement at post-operative follow-up, in line with previous comments on the benefits of surgical resection of symptomatic metastases , though it has not yet been quantified in the literature to our knowledge. This reassures that many patients will benefit symptomatically from surgery and that failure to respond to steroids prior to does not necessarily portend a poor long-term functional prognosis. While formal analysis of the patients who did not improve neurologically post-operatively was beyond the scope of this paper, many among this group succumbed to their disease within 3 to 6 months indirectly suggesting a significant disease burden and poor functional baseline.
Regardless, a response to pre-operative steroids appears to be best predictive (100% in this cohort) of durable neurologic improvement. This finding greatly improves functional prognostication and, in so doing, potentially transforms pre-treatment counseling and decision-making for this often medically and ethically complex patient population. Patients with short life expectancy, but for whom symptom palliation would significantly benefit their quality of life, would benefit from such foresight in weighing a decision to undergo surgery. However, symptom relief and restoration of neurologic functions may be equally critical to patients with well-controlled systemic disease and a favorable overall prognosis. In the immediate term, these findings also inform peri-operative management by demonstrating that some patients are unlikely to see symptomatic benefit with steroids. Given the number of potential side effects [1, 2, 20, 21] and association with poorer outcomes in glioblastoma , foregoing steroid administration in these groups could improve outcomes overall.
Our results do not support a specific dosing regimen, as no differences in symptom response rate were seen between different dosages, dosing regimens, or duration of therapy. Patients showed improvement roughly 24 hours after steroids were initiated, or after receiving 17mg cumulatively, yet there was no clear additive response to steroids with continued therapy. These values are inter-related, given a 16 mg/day regimen was most common in this population, however it does raise questions of continuing versus tapering steroids beyond this time point or dosage in patients who have not improved. Remarkably few studies have pursued the question of optimal steroid regimen, with the best evidence from Vecht, et. al  suggesting a lower dosage of 4 or 8 mg daily yielded equivalent performance status to 16 mg daily in patients awaiting radiation therapy. Further prospective study comparing dosing regimens is merited and necessary in better tailoring peri-operative steroid use for these patients.
Further work is needed to determine the mechanism behind these results. Previous imaging studies have shown a decrease in peri-tumoral edema [5-8] and a restoration of CBF  in these areas following steroid administration, but the symptomatic benefits associated with these effects remain uncharacterized. Presumably, successful tumor resection would permanently resolve local edema and normalize CBF, allowing for restoration of local neurologic functions. Instances where focal neurologic impairment is due to direct tumor disruption of eloquent pathways, rather than infiltrating edema, may be less likely to benefit from steroids however, as has previously been reported . Correlating imaging changes with steroid administration would be informative in further guiding steroid indications and predicting outcomes after therapy.
This study has the usual limitations of a retrospective review. Specifically, inter-observer variability could confound assessments of improvement or stability in a patient’s neurologic exam. Because of this, deficits felt to be most objectively documented such as language, visual field, and motor domains were included. High inter-observer agreement has previously been found in strength and language testing in rigorous study . Altered mental status and sensory hemineglect were present frequently in this patient population, but were not included for analysis because of variability in documentation between examiners. Tumor location and neurologic deficit were not correlated in detail, as this was felt to be beyond the scope of this study and difficult to propose without consistent functional imaging or intraoperative mapping data. In any case, a deficit was only included if it corresponded to a tumor on the contralateral hemisphere. For this reason as well, deficits such as dysmetria were not analyzed as part of this report given their dubious relation to varied tumor locations. Unfortunately, cohort size limited multivariate analysis by deficit type.