Growth Curves of Subependymal Giant Cell Tumors in Tuberous Sclerosis Complex

Standard Protocol Approvals and Inclusion Criteria

This study was approved by the institutional review board. Patient data were retrospectively reviewed and de-identified, and storage was encrypted and password-protected. The need for written consent was waived.

We retrospectively reviewed the clinical and imaging medical records of patients with TSC seen at our institution between 2002 and 2018.

Inclusion criteria were the following: 1) confirmed diagnosis of TSC by consensus criteria⁸ or by genetic testing; 2) at least 3 serial MRIs, including 3D Brain Volume Imaging (BRAVO; GE Healthcare), MPRAGE, and spoiled gradient echo; 3) earliest MR imaging obtained before 20 years of age; and 4) radiologic evidence of SGCTs and/or SENs. For all patients who eventually underwent surgical resection for an SGCT, at least 3 preoperative MRIs had to span at least a 1.5- year surveillance period. For the non-SGCT group, at least 3 MRIs had to span at least a 5-year surveillance period; a longer surveillance period was beneficial to ensure that the nodules did not eventually grow into SGCTs. Exclusion criteria were the following: 1) treatment with mechanistic target of rapamycin inhibitors during the surveillance period, 2) surgical resection of an SGCT before the surveillance period, and 3) nondiagnostic MRIs.

Clinical Data

The clinical records were reviewed for age at earliest MR imaging, sex, genetic testing, and age at surgical resection.

Imaging Data

All nodules at the FOM, small lesions, indeterminate lesions, and resected lesions, were measured from all available brain MRIs. A single researcher (J.R.B.) performed all measurements. The senior neuroradiologist (P.C.) with 17 years' postfellowship experience trained a researcher (J.R.B.) on an initial training data set of 20 MRIs. The research assistant then performed all remaining measurements, and for certain cases when they were in question, the measurements were reviewed and confirmed for accuracy by P.C. The senior neuroradiologist measured 10 nodules in 10 patients independently, and the interreader reliability was calculated using the intraclass correlation coefficient. The intraclass correlation coefficients for anterior-posterior (AP), superior-inferior (SI), and transverse (TV) measurements were 0.94, 0.97, and 0.93, respectively.

All measurements were made using the available T1 volumetric sequence for each MR imaging study date with a BRAVO, MPRAGE, or echo-spoiled gradient echo sequence. The T1 volumetric sequence was brought up in a 3D viewer in the institutional PACS, and the axial, coronal, and sagittal reformats were simultaneously displayed. The AP, TV, and SI measurements were performed.

Imaging Parameters

Parameters for the volumetric T1 sequences were as follows: For the echo-spoiled gradient echo sequence (n = 110), the parameters were section thickness = 1 mm, gap = 0 mm, TR = 74 ms, TE = 7 ms, phase = 190–220, frequency = 190–220, number of excitations 1 and FOV = 190–220 × 190–220 mm. For the BRAVO sequence (n = 1), the parameters were section thickness = 1 mm, gap = 0 mm, TR = 74 ms, TE = 6.508 ms, phase = 260, frequency = 260, number of excitations 1 and FOV = 260 mm. For the MPRAGE sequence (n = 160), the parameters were thickness = 1 mm, gap = 0 mm, TR = 2530 ms, TE = 2530 ms, phase = 192–256, frequency = 192–256, number of excitations 1 and FOV = 192–256 × 192–256 mm.

Lesion Categorization

Lesions were retrospectively categorized using the measurements obtained on the most recent MR imaging during the surveillance period. Small lesions were defined as lesions <1 cm in maximum diameter and may be considered terminologically per the consensus guidelines as synonymous with SENs.⁶ Indeterminate lesions were defined as lesions >1 cm in maximum diameter but that never required surgical resection during the surveillance period or at any time point according to the medical records; such lesions are traditionally termed SGCTs.⁶ Resected lesions were defined as lesions of >1 cm in maximum diameter, which were surgically resected for an indication of obstructive hydrocephalus, and this term may be considered as synonymous with resected SGCTs.⁶

Volume Calculation and Evaluation

Volume was calculated assuming that the lesion was ellipsoid: [Volume = 4/3π (A/2)(B/2)(C/2)], where A, B, and C are the maximum dimensions of the lesion measured in 3 perpendicular planes. The volumes of the nodules were calculated on every MR imaging measured during the surveillance period.

Growth Evaluation

The available MRIs for each patient were grouped into 4 age periods: 0–5, 6–10, 11–15, and 16–20 years of age. The first and last volume measurements were taken from each age period, and the following equation was applied to determine growth rate: (Volume at Most Recent Scan – Volume at Earliest Scan) / (Age at Most Recent Scan – Age at Earliest Scan) = Growth Velocity (mm³/year). The median growth velocity and interquartile range (IQR) for each age period were calculated. This approach was then subdivided for each type of lesion (small lesions, indeterminate lesions, and resected lesions). Additionally, the percentage of lesions with growth velocity >4.2 mm³/year in a given age period was calculated; 4.2 mm³ was chosen to reflect an increase in growth of 2 mm in the AP, TV, and SI dimensions using an ellipsoid volume formula.

Finally, growth velocity was found between each MR imaging for small lesions, indeterminate lesions, and resected lesions using the equation (Volume at MR Imaging – Volume of Previous MR Imaging) / (Age at MR Imaging – Age at Previous MR Imaging) = Growth Velocity (mm³/Year). Acceleration of growth was then calculated using the formula (Growth Velocity – Previous Growth Velocity) / (Age at MR Imaging – Age at Previous MR Imaging) = Acceleration of Growth (mm³/year²). The median and IQR for growth velocity and acceleration were calculated.

Statistical Analysis

Statistical analysis was conducted using GraphPad Prism software, Version 9 (GraphPad Software). All statistical analyses were conducted with an α of .05. Normality was tested by the D'Agostino-Pearson test. Nonparametric tests were used to compare data that were determined to have a non-normal distribution, and for this data medians with IQRs were reported.

A 2-tailed Mann-Whitney U test was used to evaluate volumes of resected lesions and indeterminate lesions at baseline and the most recent scan. A Kruskal-Wallis test was used to compare growth velocity of all lesions, small lesions, indeterminate lesions, and resected lesions at 0–5, 6–10, 11–15, and 16–20 years of age. Additionally, the following comparisons were made within each age period using a 2-tailed Mann-Whitney U test with a Bonferroni correction: 1) resected lesions versus indeterminate lesions, 2) resected lesions versus small lesions, 3) small lesions versus indeterminate lesions, and 4) small lesions and indeterminate lesions versus resected lesions. For the 0–5, 6–10, and 11–15 years of age periods, 4 comparisons were made. Each P value calculated within these periods was multiplied by 4. A comparison of growth velocity and acceleration on serial imaging was made using a 2-tailed Mann-Whitney U test with a Bonferroni correction to compare all lesion groups. Three comparisons were made for acceleration and velocity. Each P value was multiplied by 3. Finally, a mixed-effects analysis was performed to analyze the interactions between lesion type and age period (0–5, 6–10, 11–15, and 16–20 years of age) on growth velocity. Growth velocity was defined as the dependent variable, and lesion type and age period (0–5, 6–10, 11–15, and 16–20 years) were independent variables with an interaction term between them. Within-subject repeated measurements were accounted for using subject-level random effects.