Evaluating Neurocognitive Recovery Following Stereotactic Radiosurgery and Whole Brain Radiation Therapy: Insights from a Pooled Analysis of Three Phase III Trials

Purpose/Objective(s)

Neurocognitive changes following brain radiation therapy are significant concerns for patients (pts) with brain metastases (BM). Despite reductions in the rates of neurocognitive failure (NCF) with conformal radiation techniques such as stereotactic radiosurgery (SRS) and hippocampal-avoidance whole brain radiation therapy (HA-WBRT), a significant number of pts still experience NCF. Although there are data describing the onset and incidence of NCF, the long-term neurocognitive changes and potential functional recovery that pts experience after NCF have not been well described. Thus, we aimed to evaluate cognitive recovery (CR) following initial NCF in patients treated with SRS or WBRT by analyzing cognitive testing results from cooperative group trials.

Materials/Methods

Using the NCTN data archive, we conducted a pooled analysis of three phase III randomized clinical trials - NCCTG N107C/CEC.3 (comparing postoperative SRS vs. WBRT), NCCTG N0574 (comparing SRS vs. SRS+WBRT), and NRG Oncology CC001 (comparing HA-WBRT vs. WBRT) - and included pts who experienced NCF as predefined in each trial. Full CR was defined as pts no longer exhibiting a 1 or more standard deviation (SD) decline from baseline on any cognitive test, while recovery on individual tests was defined as at least a 1 SD improvement on a previously failed cognitive test. To estimate the incidence of CR, we used cumulative incidence function and Gray’s test. We analyzed prognostic variables associated with CR using multivariable Cox proportional hazards modeling.

Results

Two hundred eighty-eight pts who experienced trial-defined NCF were included. The pooled cumulative incidence of full CR was 38% and 42% at 6- and 12-months after onset of NCF, respectively. The incidence rates of improvement on any previously failed cognitive test were 73% and 76% for the same time points. Cumulative incidence of full CR was significantly greater with postoperative SRS vs. WBRT (HR = 2.68, Gray’s P = 0.002), as well as with SRS alone vs. SRS+WBRT (HR = 2.35, P = 0.008). There was a trend towards higher incidence of CR with HA-WBRT vs. WBRT (HR = 1.57, P = 0.059). There was no difference in rates of improvement in cognitive tests based on treatment. On multivariable pooled analysis, SRS was predictive of CR vs. WBRT (HR = 2.42, P < 0.0001). HA-WBRT demonstrated near significant association with CR vs. WBRT (HR = 1.56, P = 0.06). Age and performance status were not prognostic for CR.

Conclusion

Our analysis reveals that a sizeable proportion of pts who experience NCF following brain radiation therapy eventually demonstrate recovery. The use of conformal radiation techniques such as SRS and HA-WBRT result in greater rates of functional recovery. These findings may help counsel pts about the likelihood of meaningful neurocognitive improvement, underscoring that neurocognitive decline is not necessarily permanent. This has implications for clinical trial design and for the management of pts, highlighting the potential for recovery and adjustment in treatment strategies.