Neurocognitive function after WBRT plus SRS or SRS alone

We congratulate Chang and colleagues1 for doing a randomised trial with a neurocognitive endpoint comparing whole-brain radiotherapy (WBRT) plus stereotactic radiosurgery (SRS) with SRS alone in patients with brain tumours. Prospective trials in this setting are as challenging as they are needed. However, we feel it is necessary to highlight limitations that prevent it from being the definitive word on this topic.

The primary endpoint of neurocognitive function was assessed at a single time point of 4 months. A prospective trial (PCI-P120-9801) has shown that for patients with a survival of 4 months or less following whole-brain radiotherapy (WBRT), there is a strong correlation between decline in mean score for memory recall and time. For long-term survivors, no such correlation after WBRT was found. Furthermore, a biphasic pattern of post-WBRT neurocognitive function was noted. With multiple time points measured, neurocognitive function of long-term survivors typically declines at about the 2—4 month point, but subsequently rebounds.2 In the current study1, the SRS plus WBRT group had a median survival of 5·7 months, and therefore were expected to do poorly in tests of neurocognitive function at the 4-month point.

Another concern is the balance of the study groups. Several findings point to patients in the combined group having a disproportionately worse prognosis at outset. The SRS alone group contained an excess of women, single metastasis, recursive partioning analysis class I patients, and an absence of patients with lung and abdominal metastases. Furthermore, the combined therapy group had a greater burden in terms of disease volume. It is therefore unsurprising that baseline neurocognitive function was worse in this group. Though patients were well matched in terms of numbers of lesions, there was a disproportionately higher volume of disease in the combined therapy group. Baseline neurocognitive function is highly correlated with volume of indicator lesions, but not with number of metastases.3

Neurocognitive function at baseline has also been shown to be predictive of overall survival in patients with brain metastases.3 The comparison of two groups well matched in terms of prognostic baseline characteristics is the basis of a valid randomised trial. Though randomised, given the small numbers in this study, meaningful imbalance is not likely to be able to be statistically proven.

Of note is the high rate of surgical salvage in the SRS alone group, with 12 craniotomies done for 10 patients versus none in the WBRT group. This suggests that treatment assignment led to bias in the subsequent aggressive approach to salvage therapy in the SRS alone group. Chemotherapy was administered to more patients in the SRS alone group and for a longer duration. These factors, together with the greater burden of visceral metastases in the combined treatment arm, offer potential reasons for an excess of systemic deaths in this group.

The risks and benefits of withholding WBRT in the setting of oligometastasis is a question now at the forefront of the minds of clinical decision makers. Studies comparing the relative neurocognitive effects of combined therapy versus SRS alone are relevant, if difficult to do. Prospective trials are a welcome contribution to clinical decision making. However, in consideration of the issues above, the claim that this study represents level I evidence should be viewed with caution.

The authors declared no conflicts of interest.