Sir,

We would like to thank you for an opportunity to respond to the comments from Drs Ellsworth and Grossman in their letter to the editor concerning our recent paper, ‘Dexamethasone Exerts Profound Immunologic Interference on Treatment Efficacy for Recurrent Glioblastoma’, by Wong et al (2015).

Contrary to the assertion by the authors, our paper did not claim that the effects of dexamethasone were mediated via steroid-induced lymphopenia. It is widely accepted that dexamethasone exerts pleotropic effects on the immune system that lead to the suppression of multiple effector systems required for therapy-induced tumor rejection (Fauci, 1976; Benedetti et al, 2003). Within our single institution patient cohort, we aggressively weaned dexamethasone doses and we found that patient outcome correlated with T-cell counts. T-cell count was used as a marker of potential immunological competency to test if it correlated with outcome, as suggested by our initial observation in the phase III trial that high dexamethasone dose was correlated with a poorer survival. As pointed out by Drs Ellsworth and Grossman, the observed lymphocyte counts in our single institution cohort were probably related to patient treatment history, intrinsic immune state or both, but not necessarily to corticosteroid usage. Furthermore, overall survival as a function of the effect of dexamethasone in each of the two arms in the phase III trial was very likely independent of the T-lymphocyte counts of patients entering the trial, as supported by our single institution patient cohort where no correlation was observed between dexamethasone dose and T-lymphocyte count.

The authors also cited their work on the immunosuppressive effect of radiation and temozolomide when given to patients with newly diagnosed glioblastomas (Grossman et al, 2011). They found that 40% of patients had <200 CD4 cells mm−3 2 months after initiation of treatment and this was associated with a poorer survival when compared with those with ⩾200 CD4 cells mm−3. Given that corticosteroid use was not a controlled variable, it is possible that dexamethasone may have contributed to the poor survival outcome in this study. Regardless, the overall conclusion of their study was also consistent with our utilisation of T-lymphocyte counts as a marker of poor outcome. Furthermore, an earlier study by Hughes et al (2005) investigated the phenomenon of lymphopenia in the pre-temozolomide chemo-irradiation era and found that 24% of the cohort had <200 CD4 cells mm−3 whereas 76% had ⩾200 CD4 cells mm−3. Therefore, it is possible that the addition of temozolomide to dexamethasone plus radiotherapy increased the proportion of patients who developed poor outcome and low CD4 lymphocyte count (from 24 to 40%). Taken together, it may be important to re-examine the potential role of dexamethasone in these two studies.

Lastly, the authors also cited that treatment-related lymphopenia is a marker of poor outcome in pancreatic and non-small cell lung cancers (Balmanoukian et al, 2012; Campian et al, 2013; Tang et al, 2014; Wild et al, 2015). Our data are consistent with this contention, but do not address the cause of the low T-lymphocyte counts in our patients. It is notable that patients in these studies also received concurrent emetogenic chemotherapies, such as taxol/carboplatin, gemcitabine or gemcitabine/carboplatin, and dexamethasone was likely an important antiemetic in the premedication regimen and may therefore confound the outcome analysis.

Although it is hard to absolutely devolve the contribution of dexamethasone from prior radiation and chemotherapy effects in patients with recurrent glioblastoma, the NovoTTF-100A monotherapy arm in the phase III trial nevertheless offered us a unique opportunity to evaluate the sole effect of dexamethasone dosage because the influence of prior radiation and chemotherapy was randomized and balanced. In contrast to commonly used chemotherapeutic regimens (Grossman et al, 2011), NovoTTF-100A does not exert such deleterious effects on the immune system. Given these conditions, we were able to determine that subjects who received a dexamethasone dose of ⩾4.1 mg day−1 had a significantly shorter survival than those who took <4.1 mg day−1. Therefore, one of the obvious implications of our work is that future clinical trials in the glioblastoma population may need to control for the confounding dexamethasone effect in outcome. Furthermore, it may be worthwhile to re-examine treatment outcomes of prior clinical trials based on dexamethasone stratification.