Histopathological correlates with survival in reoperated glioblastomas

Abstract

The addition of concomitant and adjuvant chemotherapy to radiation therapy after surgical resection has increased significantly the survival of patients with glioblastoma (GB). In conjunction, there has been an increasing fraction of patients who present with new enlarged areas of contrast enhancement and edema on post-treatment imaging that improve without further treatment. It remains to be established how this phenomenon, commonly termed pseudoprogression, can be distinguished from true tumor recurrence defined as the histological presence of active high-grade tumor, as well as its prognostic significance. Data for over 500 patients undergoing surgery for recurrent GB were reviewed. Pathological specimens were categorized as those that contained active high-grade glioma in any amount, and those that did not. Patient survival was compared between these two groups, and independent associations were assessed using Cox proportionate hazards regression analysis. 59 patients met the study criteria including complete pathological and follow-up data. Mean age was 53 ± 11 years. Median survival from suspected recurrence and initial diagnosis were 8 [5-14] and 20 [12-30] months. Seventeen patients (29 %) had no evidence of active high-grade tumor and 42 (71 %) had at least focal active high-grade glioma. Pathologic pseudoprogression at re-operation (p = 0.03) and gross total resection (p = 0.01) were independently associated with survival. The histopathological features defined here and used to assess the tumor at reoperation were independently associated with survival. These findings may be important in designing treatment strategies and clinical trial endpoints for patients with GB.

Fig. 1

Histological features of active high grade tumor—true tumor recurrence. a Highly cellular tumor, in this case around an area of necrosis, establishes the presence of “active” high grade tumor. While common in pretreatment specimens, perinecrotic pseudopalisading was, as here, usually absent. b Microvascular proliferation and highly cellular tumor tissue establish the presence of active recurrent glioblastoma. c While cytologically bland, small tumor cells in this density were considered evidence of active recurrent tumor. Mitoses were typically present (arrow) but were sparse or even absent in small specimens. Hypertrophic endothelial cells are consistent with, but not diagnostic of, active high grade tumor

Fig. 2

Histological features of treatment effect—pseudoprogression. a. Necrosis, left, hyalinized vessels, and pleomorphic tumor cells in low density are considered evidence of treatment effect, not expressions of recurrent or residual active high grade tumor. b Areas of brightly eosinophilic coagulation necrosis are common effects of treatment. Telangiectatic vessels (arrows) are additional, but less specific, features. c Necrosis in this treated lesion is associated with fibrillary and gemistocytic astrocytes in low density. It can be difficult in cases such as this to determine whether these cells are reactive or neoplastic. Even if the latter, we did consider them evidence of tumor recurrence. d Well differentiated neoplastic astrocytes in low numbers were not interpreted as evidence of recurrent active, high grade glioma. Coagulation necrosis in a paucicellular lesion such as this is typical of treatment effect. e Widely scattered pleomorphic astrocytes consistent with irradiated tumor cells were not taken as evidence of active high grade astrocytoma. f Fibrinoid necrosis of vessels is a common feature of treatment effect

Fig. 3

Stepwise multivariate Cox proportionate hazards regression analysis with Forest plot demonstrating independent association with survival from re-operation. (<0.15 to enter, ≥0.05 to exit)

Fig. 4

Kaplan–Meier survival plot by recurrence pathology from (A) time of suspected recurrence (Log rank p = 0.03) (solid line pseudoprogression, dotted line active GB recurrence). Patients without active tumor at the time of suspected recurrence went on to live significantly longer than those patients with active high grade tumor recurrence (p = 0.03). This was the case even without additional treatments until the time of active tumor recurrence