https://orcid.org/0000-0002-1254-5310
Progression of glioblastoma inevitably occurs, and re-resection of recurrent glioblastoma represents a viable salvage therapy. Here, re-resection translates into more favorable outcomes when minimal residual contrast-enhancing (CE) tumor is achieved.1,2 Whether patterns of recurrence affect the maximal safe extent of re-resection and therefore overall outcome is unclear. To delineate associations between recurrence patterns and postoperative outcomes, we retrospectively collected clinical data from patients with first recurrence from a previously resected IDH-wild-type glioblastoma undergoing re-resection at 8 neuro-oncological centers in Europe and the United States.1,3 Tumor volumetrics were determined, and patients were stratified according to recurrence patterns on preoperative imaging (Figure 1A).
Recurrence patterns and outcome in glioblastomas undergoing re-resection. (A) Patterns of first recurrence for patients with a previously resected glioblastoma. (B) Initial tumor location (“dominant”: dominant hemisphere), time to first recurrence, and pre- as well as postoperative CE tumor volumes stratified according to the recurrence patterns. (C) Postoperative Karnofsky Performance Status and non-surgical therapies following re-resection stratified according to the recurrence patterns. (D) Kaplan–Meier estimates of overall survival after re-resection for patients with progressive residual CE tumor (n = 50), recurrence bordering the resection cavity (n = 187), distant recurrences (n = 45), and multiple recurrence patterns (n = 27). Points indicate deceased or censored patients, light shading indicates SEM.
Among 309 patients with a previously resected glioblastoma, the median time to first recurrence was 10 ± 0.4 months. Recurrences were either progression of residual CE tumor (50 patients, 16.1%), bordered the prior resection cavity (187 patients, 60.5%), distant recurrences (45 patients, 14.6%), or synchronous occurrence of more than 1 recurrence pattern (27 patients, 8.7%) (Figure 1B). With a median time to first progression of 6 ± 0.7 months, progression of residual CE tumors occurred earlier than the other recurrence patterns (P = .001 comparing all groups; “residual CE progression” vs. “distant recurrence”: HR 1.86, CI 1.2–2.8). At re-resection, the lowest postoperative CE tumor volumes (0.2 ± 0.1 cm3) were achieved in distant recurrences while higher postoperative tumor volumes were found in progressive CE tumors (1.9 ± 0.5 cm3), recurrences bordering the resection cavity (2.0 ± 0.3 cm3), or multiple recurrence patterns (2.5 ± 0.8 cm3) (P = .001). This might be due to less eloquent structures allowing more extensive resection in distant recurrences as characterized by a lower frequency of the dominant hemisphere being involved (19/45 patients, 42.2%). In line with this assumption, the highest postoperative Karnofsky performance status scores (KPS) were found in patients with distant recurrences (median KPS: 90 ± 2.3; P = .013) while there was no difference in non-surgical second-line therapies between patients with different recurrence patterns (Figure 1C). Median overall survival following re-resection was 10 ± 0.6 months, and the better clinical status and low postoperative CE tumor volumes of distant recurrences together with potential biological differences might have translated into the most favorable outcome (median overall survival: 16 ± 3.2 months; P = .017 comparing all groups) (Figure 1D). In turn, the least favorable outcome was observed for patients with multiple recurrence patterns who also had the highest postoperative CE tumor and the lowest KPS scores (median overall survival: 8 ± 1.3 months; “multiple patterns” vs. “distant recurrence”: HR 2.15, CI 1.2–3.8).
We provide evidence that among recurrent glioblastomas undergoing re-resection, different recurrence patterns are associated with distinct survival outcomes due to distinct accessibility for extensive resection. Although distant recurrences reflect the capability of glioblastoma cells to migrate throughout the brain,4,5 our findings indicate that re-resection appears reasonable when pre-surgical planning yields stable disease around the initial resection cavity and minimal residual CE tumor following re-resection is assumed. As such, the outcome of distance recurrences was surprisingly favorable. It remains speculative whether beneficial biological properties of distant recurrences outside the previous radiation field contributed to the survival differences. In contrast, re-resection of recurrent disease in proximity to a heavily pretreated resection cavity might be limited due to bordering eloquent structures (that also limited the degree of first resection) which dictate the similar postoperative prognosis between patients with progressive CE tumors and newly occurring recurrences at the border of the resection cavity. The shorter time to first progression for progressive CE tumor underlines the importance of achieving maximal resection at first resection.6 As we included patients undergoing re-resection, we cannot comment on a comparison between re-resection and non-surgical therapies for each recurrence pattern. The current study points towards a more favorable outcome when maximal resection of CE disease is achieved during the initial resection and re-resection.
Funding
None declared.
Conflict of interest statement
M.W.—Research grants: Quercis, Versameb. Honoraria or advisory board participation and consulting: Bayer, Curevac, Medac, Neurosense, Novartis, Novocure, Orbus, Philogen, Roche, Servier. M.v.d.B.—Consultant: Servier, Boehringer Ingelheim, Carthera, Nerviano, Genenta, Astra Zeneca, Fore Biotherapeutic, Incyte, Chimerix, Roche. N.T.—Research grant: Medtronic. Founder: BrainDynamics; Advisory Board: Nervonik, BrainGrade. R.R.—Honoraria, advisory board, and consulting: UCB, Bayer, Novocure, Genenta, Servier. M.A.V.—Indirect equity and patent royalty interests: Infuseon Therapeutics. Honoraria: Chimerix, Midatech. Research grants: DeNovo Pharma, Oncosynergy, Infuseon, Chimerix. J.-C.T.—Research grants: Novocure, Munich Surgical Imaging. Advisory board: AAA Novartis. Royalties: Springer Publisher. P.K.; A.D.; J.S.Y.; S.T.J.; N.Te.; L.H.; T.S.; C.Y.M.; F.B.; L.N.; R.A.M.; A.F.H.; N.Th.; J.B.; S.H.J.; A.M.M.; L.B.; O.S.; S.J.G.; S.M.C.; M.S.B.; Y.E.—None.
Author Contributions
Study concept and design: P.K., J.C.T. Data collection: P.K., A.D., J.S.Y., S.T.J., N.Te., L.H., T.S., C.Y.M., F.B., L.N., R.A.M., A.F.H. Data analysis and interpretation: P.K., A.D., J.S.Y., L.H., M.W., A.M.M., R.R., M.A.V., L.B., O.S., S.J.G., S.M.C., M.S.B., Y.E., J.C.T. Statistics: P.K., J.C.T. Manuscript drafting: P.K., J.C.T. Manuscript revising: P.K., A.D., J.S.Y., S.T.J., N.Te., L.H., T.S., C.Y.M., F.B., L.N., R.A.M., A.F.H., M.W., M.v.d.B., N.Th., J.B., S.H.J., A.M.M., N.Ta., R.R., M.A.V., L.B., O.S., S.J.G., S.M.C., M.S.B., J.C.T.
Acknowledgments
The authors thank all patients who contributed to the results of the study. Parts of Figure 1A have been created with the help of BioRender.com.
