Malignant transformation (MT) of low-grade gliomas (LGGs) to a higher-grade variant seems inevitable, yet it remains unclear which LGG patients will progress to grade 3 or even directly to grade 4 after receiving a long course of treatment. To elucidate this, we conducted a retrospective cohort study based on 229 adults with recurrent LGG. Our study aimed to disclose the characteristics of different MT patterns and to build predictive models for patients with LGG. Patients were allocated into group 2-2 (n = 81, 35.4%), group 2-3 (n = 91, 39.7%), and group 2-4 (n = 57, 24.9%), based on their MT patterns. Patients who underwent MT showed lower Karnofsky performance scale (KPS) scores, larger tumor sizes, smaller extents of resection (EOR), higher Ki-67 indices, lower rates of 1p/19q codeletion, but higher rates of subventricular involvement, radiotherapy, chemotherapy, astrocytoma, and post-progression enhancement (PPE) compared with those in group 2-2 (p < 0.01). On multivariate logistic regression, 1p/19q codeletion, Ki-67 index, radiotherapy, EOR, and KPS score were independently associated with MT (p < 0.05). Survival analyses demonstrated that patients in group 2-2 had the longest survival, followed by group 2-3 and then group 2-4 (p < 0.0001). Based on these independent parameters, we constructed a nomogram model that exhibited superior potential (sensitivity: 0.864, specificity: 0.814, and accuracy: 0.843) compared with PPE in early prediction of MT. Combining the factors of 1p/19q codeletion, Ki-67 index, radiotherapy, EOR, and KPS score that were presented at initial diagnosis could precisely forecast the subsequent MT patterns of patients with LGG.
Keywords: low-grade gliomas; malignant transformation; mathematical model; predictive marker; prognosis.
© 2023 The Authors. Cancer Science published by John Wiley & Sons Australia, Ltd on behalf of Japanese Cancer Association.