Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

ADC Histogram Analysis of Pediatric Low-Grade Glioma Treated with Selumetinib: A Report from the Pediatric Brain Tumor Consortium

ADC Histogram Analysis of Pediatric Low-Grade Glioma Treated with Selumetinib: A Report from the... ORIGINAL RESEARCH PEDIATRICS ADC Histogram Analysis of Pediatric Low-Grade Glioma Treated with Selumetinib: A Report from the Pediatric Brain Tumor Consortium S. Vajapeyam, D. Brown, A. Ziaei, S. Wu, G. Vezina, J.S. Stern, A. Panigrahy, Z. Patay, B. Tamrazi, J.Y. Jones, S.S. Haque, D.S. Enterline, S. Cha, B.V. Jones, K.W. Yeom, A. Onar-Thomas, I.J. Dunkel, M. Fouladi, J.R. Fangusaro, and T.Y. Poussaint ABSTRACT BACKGROUND AND PURPOSE: Selumetinib is a promising MAP (mitogen-activated protein) kinase (MEK) 1/2 inhibitor treatment for pediatric low-grade gliomas. We hypothesized that MR imaging–derived ADC histogram metrics would be associated with survival and response to treatment with selumetinib. MATERIALS AND METHODS: Children with recurrent, refractory, or progressive pediatric low-grade gliomas who had World Health Organization grade I pilocytic astrocytoma with KIAA1549–BRAF fusion or the BRAF V600E mutation (stratum 1), neurofibromatosis type 1–associated pediatric low-grade gliomas (stratum 3), or sporadic non-neurofibromatosis type 1 optic pathway and hypothala- mic glioma (OPHG) (stratum 4) were treated with selumetinib for up to 2 years. Quantitative ADC histogram metrics were analyzed for total and enhancing tumor volumes at baseline and during treatment. RESULTS: Each stratum comprised 25 patients. Stratum 1 responders showed lower values of SD of baseline ADC_total as well as a larger decrease with time on treatment in ADC_total mean, mode, and median compared with nonresponders. Stratum 3 responders showed a greater longitudinal decrease in ADC_total. In stratum 4, higher baseline ADC_total skewness and kurtosis were associated with shorter pro- gression-free survival. When all 3 strata were combined, responders showed a greater decrease with time in ADC_total mode and median. Compared with sporadic OPHG, neurofibromatosis type 1–associated OPHG had lower values of ADC_total mean, mode, and median as well as ADC_enhancement mean and median and higher values of ADC_total skewness and kurtosis at baseline. The longitudinal decrease in ADC_total median during treatment was significantly greater in sporadicOPHG comparedwithneurofibromatosis type 1–associated OPHG. CONCLUSIONS: ADC histogram metrics are associated with progression-free survival and response to treatment with selumetinib in pediatric low-grade gliomas. ABBREVIATIONS: MEK ¼ MAP (mitogen-activated protein) kinase; NF1 ¼ neurofibromatosis type 1; OPHG ¼ optic pathway and hypothalamic glioma; PBTC ¼ Pediatric Brain Tumor Consortium; PFS ¼ progression-free survival; pLGG ¼ pediatric low-grade glioma; WHO ¼ World Health Organization ediatric low-grade gliomas (pLGGs) are the most commonly (WHO) grade I pilocytic astrocytoma is the most frequent pri- occurring childhood brain tumor and comprise 40%–50% mary brain tumor in individuals 0–19 years of age, accounting for 1,2 of all childhood CNS tumors. World Health Organization 15% of children’s and adolescents’ (0–19 years) and 17.8% of Department of Radiology (K.W.Y.), Stanford University School of Medicine, Received August 10, 2021; accepted after revision January 1, 2022. Stanford, California; and Department of Hematology, Oncology, and Stem Cell From the Department of Radiology (S.V., T.Y.P.), Boston Children’sHospital, Harvard Transplantation (J.R.F.), Children’s Healthcare of Atlanta and Emory University, Medical School, Boston, Massachusetts; Department of Radiology (D.B.), Massachusetts Atlanta, Georgia. General Hospital, Boston, Massachusetts; Department of Radiology (A.Z.), Boston This work was supported by the American Lebanese Syrian Associated Charities, Children’s Hospital, Boston, Massachusetts; Department of Biostatistics (S.W., A.O.-T.), AstraZeneca, US Department of Health and Human Services, National Institutes of St Jude Children’s Research Hospital, Memphis, Tennessee; Department of Radiology Health, National Cancer Institute, National Cancer Institute Cancer Therapy (G.V.), Children's National Medical Center, Washington, DC; Department of Radiology Evaluation Program, National Institutes of Health/National Cancer Institute Cancer (J.S.S.), Ann and Robert H Lurie Children's Hospital of Chicago, Chicago, Illinois; Center Support Grant P30 CA008748, UM1CA081457. Department of Radiology (A.P.), Children's Hospital of Pittsburgh, Pittsburgh, Pennsylvania; Department of Diagnostic Imaging (Z.P.), St Jude Children's Research Please address correspondence to Sridhar Vajapeyam, MD, Children’s Hospital Hospital,Memphis,Tennessee;Department of Radiology (B.T.), Children's Hospital Los Boston, Radiology, 300 Longwood Ave, Boston, MA 02115; e-mail: Angeles, Los Angeles, California; Department of Radiology (J.Y.J., M.F.), Nationwide sridhar.vajapeyam@childrens.harvard.edu Children's Hospital, Columbus, Ohio; Department of Radiology (S.S.H., I.J.D.), Memorial Indicates open access to non-subscribers at www.ajnr.org Sloan Kettering Cancer Center, New York, New York; Department of Radiology (D.S.E.), Duke University School of Medicine, Durham, North Carolina; Department of Radiology Indicates article with online supplemental data. (S.C.), University of California San Francisco, San Francisco, California; Department of Radiology (B.V.J.), Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio; http://dx.doi.org/10.3174/ajnr.A7433 AJNR Am J Neuroradiol 43:455–61 Mar 2022 www.ajnr.org 455 1 primary childhood (0–14 years) brain tumors. pLGGs are bio- Imaging and Image Analysis logically distinct from their adult counterparts. Unlike adult Standard MR Imaging Evaluation. Standard MR imaging, which low-grade gliomas that occur mostly in the cerebral hemispheres included T2-FLAIR and axial pre- and postcontrast T1-weighted and transform into higher-grade gliomas, pLGGs can occur images, as well as DTI, was performed primarily on 3T scanners throughout the central nervous system, are molecularly distinct, at the start of treatment followed by every 2 months during the 3,4 rarely undergo malignant transformation, and require different first year of therapy and then every 3 months thereafter. therapies. Additionally, high-resolution MR imaging of the orbits and optic Total surgical resection is the first-line of treatment for pLGGs pathway was performed for optic pathway tumors. and can be curative. However, this is not always feasible due to tu- Patients whose tumors achieved an MR imaging response mor location, particularly in the brain stem or optic pathway and (complete or partial response) assessed locally underwent central hypothalamic gliomas (OPHGs), and other therapies are required. radiographic review at the PBTC Neuroimaging Center. A com- Radiation therapy can be effective but may cause considerable neu- plete response was defined as complete tumor disappearance on rocognitive deficits in young children and may increase the risk of T2-FLAIR images, no new lesions, and disappearance of all developing a secondary malignancy, particularly in children with enhancement on T1 postcontrast imaging. A partial response was neurofibromatosis type 1 (NF1). Several chemotherapy regimens defined as at least 50% tumor reduction (in a 2D area calculated as have shown promise in pLGGs, with 5-year progression-free sur- a product of 2 perpendicular linear measurements) on T2-FLAIR. vival (PFS) and overall survival rates of 35%–45% and 85%–100%, Stable disease was defined as neither a sufficient increase nor a respectively, in sporadic pLGG and 5-year 60%–70% PFS and reduction to qualify as a partial response or progressive disease. 90%–100% overall survival rates in NF1-associated pLGG. Progressive disease was a .25% increase or the development of Selumetinib is a potent orally available MAP (mitogen-activated new lesions. These response criteria, while similar to the latest protein) kinase (MEK) 1/2 inhibitor that has recently shown great Response Assessment in Pediatric Neuro-Oncology recommenda- 7,8 efficacy in refractory, recurrent, and progressive pLGGs. We tions, were developed earlier by a PBTC consensus panel and explored the associations of advanced diffusion MR imaging met- used both for local and central imaging response assessments. rics with response and survival in pLGGs treated with selumetinib DTI Acquisition and ADC Histogram Analysis. DTI data were in a large prospective Phase II Pediatric Brain Tumor Consortium acquired with the following acquisition parameters on a 3T scanner: (PBTC) trial, PBTC029B. section thickness = 2.2 mm, TR = 8800 ms, TE = 88 ms, FOV = 220 mm, b-value = 1000 s/mm ,35directions. By meansof the mu- MATERIALS AND METHODS tual information algorithm in FSL (http://www.fmrib.ox.ac.uk/ Subjects fsl), ADC and postcontrast T1 volumes were registered to FLAIR This was a multicenter, National Cancer Institute–sponsored volume as described previously. 3D ROIs comprising the total tu- Phase II study conducted by the PBTC using the MEK 1/2 inhibi- mor volume from the FLAIR images and the enhancing tumor vol- tor selumetinib in patients with pediatric low-grade gliomas ume from postcontrast T1 were automatically generated using the 8,9 treated at 11 PBTC member hospitals. Patients 3–21 years of thresholding feature in Fiji (https://fiji.sc), and the corresponding age with a Lansky/Karnofsky Performance Status score of .60 ADC was used to generate the ADC_total and ADC_enhancement and the presence of recurrent, refractory, or progressive pediatric volumes, respectively. These volumes were thresholded using a uni- 6 2 low-grade glioma after at least 1 standard therapy, including form range of 600–2600  10 mm /s to automatically exclude chemotherapy and radiation therapy, were eligible for inclusion. cysts, necrosis, and hemorrhage; corresponding ADC_total and Patients were assigned to 6 unique strata according to histology, ADC_enhancement histograms were generated. 8,9 tumor location, NF1 status, and BRAF aberration status. Histogram metrics used for statistical analysis were the mean, 8,9 Clinical analyses of strata 1, 3, and 4 have been completed, and SD, mode, median, skewness, and kurtosis of these histograms at the imaging data were used in this study. Stratum 1 comprised baseline and 6, 12, 18, and 24 months into treatment and also at patients with WHO grade I pilocytic astrocytoma with either progression. In studies with multiple lesions, the primary target KIAA1549-BRAF fusion or the BRAF V600E (Val600Glu) muta- lesion was evaluated for all analyses. tion. Stratum 3 comprised patients with imaging- or biopsy- For each of the imaging metrics, the baseline value and the proved pLGGs and a clinical or genetic diagnosis of NF1.Stratum time-dependent longitudinal change in the metrics during the 4 comprised OPHG pLGGs not associated with NF1. course of treatment were examined for associations with tumor Selumetinib was provided as capsules given orally at a dose of volume, response, and PFS. 25 mg/m twice daily in 28-day courses for up to 26 courses. The primary end point was the proportion of patients with a stratum- Statistical Methods specific objective response (partial response or complete The Wilcoxon rank-sum test was used to test the differences in response), as assessed by the local site and sustained for at least ADC parameters at baseline between 2 groups. Subjects with at 8,9 8 weeks. All responses were confirmed by central review. least 1 follow-up MR imaging after baseline were eligible for The protocol was approved by the Cancer Therapy Evaluation time-dependent longitudinal analyses. Tumor volumes were cor- Program (CTEP) as well as each site’s institutional review board. related within the patients because the measures were longitudi- All patients or legal guardians provided written, informed con- nal. The mixed model was used to investigate the correlation sent when applicable based on institutional guidelines. between tumor volume and ADC metrics, taking intrapatient 456 Vajapeyam Mar 2022 www.ajnr.org significantly associated with response (P=.009, q = 0.04), with responders showing lower values than nonres- ponders. The best response achieved was used in all analyses. There were significant associations of response with longitudinal change across time on treatment in the ADC_total mean (P=.02, q = 0.05), ADC_total mode (P=.02, q = 0.07), and ADC_total me- dian (P=.03, q = 0.1), with responders showing a larger decrease in these pa- rameters across time than nonres- ponders. Five of the 8 responders showed a transient increase in ADC followed by a decrease by 24 months as seen in Fig 1. Of the 21 enhancing tumors in stratum 1, sixteen were eli- gible for longitudinal analysis and 8 of these showed a response to selumeti- nib. None of the histogram metrics of ADC_enhancement were associated with a response in the longitudinal analyses. No group differences were found in ADC histogram metrics between FIG 1. The decrease at 24 months in the mean, median, and mode of ADC_total with treatment in a 6-year-old boy with a WHO grade I glioma with a BRAF aberration who responded to selume- the BRAF fusion and the BRAF muta- tinib. The initial transient increase in these metrics seen here was observed in 5 of the 8 respond- tion groups either at baseline or in the ers in stratum 1 and is possibly linked to increased water mobility due to early cell death, longitudinal analysis. followed by tissue consolidation later in treatment. Tx indicates treatment. Stratum 3 Twenty-five patients (mean age at study entry, 10.2 years; range, 3.5–16.6 years; 15 males and 10 correlation into account. Mixed-effects models were used to esti- females) with any NF1-associated pediatric low-grade glioma mate trends across time in the ADC parameters as well as differen- (WHO grades I and II) were enrolled in stratum 3. Nine patients ces in longitudinal changes between groups. Cox models were used showed sustained partial response, 15 had stable disease, and 1 to investigate the association between PFS and ADC parameters at showed progressive disease. The 2-year PFS in stratum 3 was baseline as well as with time. The latter incorporated time-depend- 96%. These results were reported previously. ent ADC parameters. Q values were calculated within each stratum None of the baseline ADC histogram metrics were associated for PFS and response-based analyses separately to adjust for multi- with response or PFS. Longitudinal change with time on treatment plicity via the false discovery rate if.1 parameter was significant. of ADC_total mode was significantly associated with response A q-value of 10% (0.1) was considered significant. Similarly, q-val- (P = .03, q = 0.07), with responders showing a greater decrease ues were also used for cross-strata comparisons. across time than nonresponders. Five of 10 responders showed a transient increase in the ADC followed by a decrease by 24 months. RESULTS Stratum 1 Stratum 4 Twenty-five patients (mean age at study entry, 9.2 years; range, Twenty-five patients (mean age at study entry, 9.4 years; range, 3.9–20.8 years;12males and 13females) with WHO grade I pilo- 3.7–17.6 years; 12 males and 13 females) with sporadic non-NF1 cytic astrocytoma with either 1 of the 2 most common BRAF aber- pediatric optic pathway and hypothalamic low-grade gliomas rations (KIAA1549–BRAF fusion or the BRAF V600E [Val600Glu] (WHO grades I and II) were enrolled in stratum 4. Six patients mutation) were enrolled. Eighteen of these had a KIAA1549–BRAF showed sustained partial response, 14 had stable disease, and 5 fusion, and the remaining 7 had the BRAFV 600E mutation. Nine showed progressive disease. The 2-year PFS in stratum 4 was patients showed sustained partial response, 9 had stable disease, 73.8% (SD 9.3%) as reported previously. and 7 showed progressive disease. Two-year PFS in stratum 1 was At baseline, there were statistically significant associations of 70%. These results were reported previously. ADC_total skewness (P=.02, q = 0.06) and ADC_total kurtosis None of the baseline ADC_total histogram parameters were (P = .02, q = 0.06) with PFS in stratum 4. Patients in stratum 4 associated with PFS. Only the SD of ADC_total at baseline was with higher baseline skewness or kurtosis of ADC_total had AJNR Am J Neuroradiol 43:455–61 Mar 2022 www.ajnr.org 457 OPHGs from stratum 3. No difference was found in PFS between the 2 groups. Enhancing tumor volume and cyst volume at baseline were both significantly lower in the NF1-associ- ated OPHG in stratum 3 compared with stratum 4 (P, .001), while there was no difference in FLAIR tumor vol- ume between the groups. Significant differences between the groups were found in baseline values of ADC_total mean (P = .005, q = 0.01), mode (P = .005, q = 0.01), median (P = .002, q = 0.003), skew- ness (P, .001, q, 0.001), and kurto- sis (P = .01, q = 0.02). The sporadic stratum 4 cohort had higher values of ADC_total mean, mode, and median and lower values of skewness and kur- FIG 2. The longitudinal trend of median ADC of the total tumor volumes for the cohort of 25 tosis. Similarly, ADC_enhancement sporadic non-NF1 OPHGs from stratum 4 shown in red, compared with the 15 NF1-associated showed higher values in sporadic OPHGs shown in blue. Mixed-effects models were used to estimate trends with time in ADC pa- OPHG for mean (P = .045, q = 0.09) rameters as well as differences in longitudinal changes between groups. Sporadic OPHG values and median (P = .04, q = 0.09) com- are higher at baseline (P=.002, q = 0.003) and show a greater decrease with time (P=.02, q =0.06) compared with NF1-associated OPHG. pared with NF1-associated OPHG. In the longitudinal analysis of change with time on treatment, the ADC_total shorter PFS (Online Supplemental Data). None of the longitudi- median decreased significantly more in the sporadic cohort in stra- nal ADC_total histogram metrics were associated with response tum 4 compared with NF1-associated stratum 3 OPHG (P = .02, or PFS. Of the 23 enhancing tumors in stratum 4, twenty were eli- q = 0.06) as shown in Fig 2, with the ADC_total mean similarly gible for longitudinal analysis with only 4 responders in this approaching significance (P = .04, q = 0.11). group. While the entire cohort showed a decrease with time in ADC_enhancement mean (P=.02, q=0.05), mode (P=.01, DISCUSSION q = 0.04), and median (P=.008, q = 0.02), there was no difference The past decade has seen an explosion of molecular data showing between responders and nonresponders. that most pLGGs upregulate the RAS–mitogen-activated protein 15,16 kinase pathway, most commonly including BRAF fusion or Combined Strata 17,18 16 mutation of the BRAF gene and NF1 mutation. More than None of the baseline ADC histogram metrics were associated with 80% of pilocytic astrocytomas have gene alterations in some com- response or PFS when all strata were combined (n = 75). Response ponent of the MEK signaling pathway; therefore, several thera- was still significantly associated with longitudinal change with time pies have been developed to target this pathway. BRAF resistance in treatment for ADC_total mode (P=.02, q = 0.06) and and tumor progression have been reported in some therapies ADC_total median (P = .03, q = 0.09). A greater decrease in these 19,20 such as sorafenib targeting non-V600e aberrations, leading to metrics was found in patients showing a response to the study therapies targeting downstream pathway components like MEK. drug. Of the 64 enhancing tumors in the combined strata, 48 were Selumetinib is one such potent orally available MEK 1/2 inhibitor eligible for longitudinal analysis, with 19 responders. The entire 7-9 that has shown promise in pLGG. cohort showed a decrease with time for the mean (P = .01, Pilocytic astrocytoma is the most frequent primary brain tu- q = 0.04), mode (P = .009, q = 0.03), and median (P = .009, q = 0.03) mor in children and can occur anywhere in the central nervous of ADC_enhancement in the longitudinal analysis, but there was system, with the most common locations being the cerebellum no difference between responders and nonresponders. There were (40%), followed by supratentorial locations (35%), the optic path- no group differences in ADC histogram metrics of tumors that way and hypothalamus (11%), and the brain stem (9%). progressed and those that showed no progression in the 24-month Histopathologically, pilocytic astrocytoma is a tumor of low-to- study period. Total tumor volume was found to be negatively asso- moderate cellularity with compact, densely fibrillated areas rich ciated with the SD of ADC_total (P = .008). in Rosenthal fibers, as well as spongy loosely textured areas com- posed of multipolar cells that have varying degrees of mucoid Optic Pathway and Hypothalamic Glioma: NF1 versus background material, often with microcysts. Most OPHGs, Sporadic Twenty-five subjects with sporadic OPHGs in stratum 4 were including those associated with NF1, are pilocytic astrocytomas. compared with the cohort with 15 NF1-associated low-grade The pathology of these tumors makes it highly likely that the 458 Vajapeyam Mar 2022 www.ajnr.org extracellular matrix is an important contributor to the ADC of associated with shorter PFS, a finding similar to those in previous these tumors. reports in pediatric diffuse intrinsic pontine glioma (DIPG). ADC maps derived from diffusion-weighted images measure Higher skewness and kurtosis signify more homogeneous tumors the diffusivity of water molecules in tissue. In the context of brain with lower ADC and are usually associated with higher cellularity tumors, ADC is influenced by tumor cellularity and the extracel- in high-grade tumors, but in these low-grade OPHGs, they may lular matrix, as well as the presence of edema, cystic components, be indicative of a lower fraction of extracellular matrix. and necrosis. Several studies have reported an inverse relation- Longitudinal ADC_total histogram metrics were not associated ship between ADC and cellularity in a variety of pediatric brain with response or PFS. 24,25 tumors. As cellularity increases, there is less extracellular When data from all 3 strata were combined, there were no space for water diffusion, leading to a lower ADC. Jost et al associations between baseline ADC histogram metrics, and PFS or examined the ADC in a cohort of 14 NF1-associated OPHGs and response, suggesting that strata-specific analyses may be more use- ful. In a study of newly diagnosed OPHGs treated with carboplatin 13 sporadic OPHGs and found that ADC was not associated with 27 27 either NF1 status or clinical aggressiveness. Yeom et al, how- and vincristine, Yeom et al found that tumors with a higher base- ever, found that in a retrospective study of a cohort of 5 patients line mean had a shorter PFS. Our study comprised previously with NF1 and 7 with sporadic pediatric OPHG, a higher baseline treated pLGGs not limited to OPHGs, which may explain why we ADC was predictive of tumor progression and that ADC then found no baseline associations. However, a greater longitudinal declined following subsequent chemotherapy with a standard decrease in mode and median ADC_total with time in treatment combination of carboplatin and vincristine. Similarly, Hsu et al was seen in responders, similar to our findings in the individual reported that the ADC declined during the response to bevacizu- strata. Tumors with smaller total tumor volume were found to mab in a cohort of 8 patients with progressive pLGG. Whereas all have a higher SD of ADC_total, probably due to the higher pro- the above studies used the mean ADC of the tumor volume, we portion of cells on the tumor periphery in smaller tumors. When the longitudinal trends in the individual and combined used ADC histogram analyses to better characterize the ADC dis- strata are considered as a whole, a consistent overall picture tribution in the tumors. ADC histograms derived from the ADC of all voxels in a tu- emerges, suggesting that a steeper drop in ADC_total values dur- mor volume are particularly well-suited to characterize the diffu- ing treatment with selumetinib occurs in patients who respond to sion properties of the entire tumor volume in brain tumors and selumetinib. This finding confirms earlier reports in the literature 27 28 have been used to differentiate pediatric brain tumors by histo- from Yeom et al and Hsu et al, who saw similar trends in 29 30 logic type and tumor grade. ADC histogram parameters have smaller samples of patients with pLGGs undergoing chemother- also been associated with molecular subtype and response to apy. This suggests that monitoring ADC during the course of 12,32 therapy in diffuse intrinsic pontine gliomas in children. treatment may provide some clinical value in assessing the Herein, we report the use of ADC histogram metrics to identify response in pLGGs. A lower ADC has long been shown to be in- prognosis and response criteria in refractory, recurrent, or pro- dicative of higher cellularity, and an increase in ADC has been shown to be a biomarker of cell death in response to treatment in gressive pLGGs treated with selumetinib. Pilocytic astrocytomas in stratum 1 that responded to selume- high-grade gliomas in adults. Our longitudinal results linking a tinib had a smaller SD of baseline ADC_total, suggesting that decrease in ADC with response is not consistent with the oppo- more homogeneous tumors responded better. In addition, the site behavior in high-grade gliomas because these are low-grade mean, median, and mode of ADC_total all decreased more dur- gliomas with a more extensive extracellular matrix. Tumor ing the course of treatment for responders compared with non- response with selumetinib in pLGGs may be the result of a 28 27,35 responders. This decrease is similar to that found by Hsu et al decrease in extracellular space, leading to the decrease in in a small sample of recurrent or progressive pLGGs treated with ADC seen in responders in our study. The impressive response to bevacizumab. The transient increase in these ADC metrics seen selumetinib reported recently in plexiform neurofibromas, in the first few months of treatment is possibly due to increased which are known to have high extracellular matrix content, water mobility due to cell death, followed by tissue consolidation may also be due to changes in the extracellular matrix. Patients with OPHGs associated with NF1 are known to have leading to a stable decrease in these values. Patients in stratum 1 longer PFS compared with those with sporadic OPHGs. When with tumors that had the KIAA1549-BRAF fusion had a longer PFS than those with the BRAF V600E mutation, but there was no we compared the 25 sporadic OPHGs in stratum 4 with 15 NF1- difference in the response rate between the groups. However, associated OPHGs in stratum 3, no significant difference in PFS our analyses of ADC histogram metrics showed no difference was seen, possibly due to the small sample size or the relatively between the BRAF fusion and BRAF mutation either at baseline short period of time of 24 months used in this analysis. or across time, possibly due to small sample size. Enhancing tumor volume as well as the volume of cysts were sig- NF-1 associated pLGG in stratum 3 showed no association nificantly higher in sporadic OPHGs compared with NF1-associ- between baseline ADC histogram metrics and either PFS or over- ated OPHGs, whereas there was no difference in overall tumor all survival. ADC_total mode decreased more with time during volume. The baseline mean, median, and mode of ADC_total as therapy in responders in stratum 3, similar to our findings in stra- well as baseline mean and median of ADC_enhancement were tum 1. higher in sporadic OPHGs compared with NF1-associated Among the patients with sporadic OPHGs in stratum 4, OPHGs. This result may be explained by the higher incidence of higher baseline skewness and kurtosis of ADC_total were cystic components in sporadic OPHGs that we report, consistent AJNR Am J Neuroradiol 43:455–61 Mar 2022 www.ajnr.org 459 39 with previous reports. The greater decrease with time in median 8. Fangusaro J, Onar-Thomas A, Young Poussaint T, et al. Selumetinib in paediatric patients with BRAF-aberrant or neurofibromatosis ADC_total we report in sporadic OPHGs in stratum 4 compared type 1-associated recurrent, refractory, or progressive low-grade with NF1-associated OPHG in stratum 3 (Fig 2) may also be asso- glioma: a multicentre, phase 2 trial. Lancet Oncol 2019;20:1011–22 ciated with a decrease in cystic components of the tumor. CrossRef Medline Although ours was larger than prior studies in the literature, 9. Fangusaro J, Onar-Thomas A, Poussaint TY, et al. Aphase 2trial of the relatively small sample size and diverse prior treatment his- selumetinib in children with recurrent optic pathway and hypo- thalamic low-grade glioma without NF1: a Pediatric Brain Tumor tories are limitations of this study. Consortium Study. Neuro Oncol 2021;23:1777–88 CrossRef Medline Machine learning methods have shown promise in differenti- 10. Fangusaro J, Witt O, Hernáiz Driever P, et al. Response assessment ating BRAF fusion and mutation and could be useful in predict- in paediatric low-grade glioma: recommendations from the ing response to therapy. This study was an observational and Response Assessment in Pediatric Neuro-Oncology (RAPNO) exploratory analysis and is the first study, to our knowledge, to working group. Lancet Oncol 2020;21:e305–16 CrossRef Medline 11. Jenkinson M, Beckmann CF, Behrens TE, et al. FSL. Neuroimage look for any associations between treatment response and 2012;62:782–90 CrossRef Medline changes in ADC in pLGGs. Our population was heterogeneous 12. Poussaint TY, Vajapeyam S, Ricci KI, et al. Apparent diffusion coef- with respect to molecular type as well as tumor location, and ficient histogram metrics correlate with survival in diffuse intrin- response to selumetinib varies by molecular type and tumor loca- sic pontine glioma: a report from the Pediatric Brain Tumor tion. Future studies with the large homogeneous samples of each Consortium. Neuro Oncol 2016;18:725–34 CrossRef Medline 13. Schindelin J, Arganda-Carreras I, Frise E, et al. Fiji: an open-source tumor type required for artificial intelligence and machine learn- platform for biological-image analysis. Nat Methods 2012;9:676–82 ing algorithms may definitively identify ADC histogram metrics CrossRef Medline as markers to identify response to selumetinib. Future work may 14. Strimmer K. fdrtool: a versatile R package for estimating local and also include exploring associations between ADC histogram met- tail area-based false discovery rates. Bioinformatics 2008;24:1461–62 rics and standardized measures of visual outcomes in OPHGs as CrossRef Medline 15. Packer RJ, Pfister S, Bouffet E, et al. Pediatric low-grade gliomas: recommended by Response Assessment in Pediatric Neuro- 10 implications of the biologic era. Neuro Oncol 2017;19:750–61 Oncology. CrossRef Medline 16. Ryall S, Zapotocky M, Fukuoka K, et al. Integrated molecular and CONCLUSIONS clinical analysis of 1,000 pediatric low-grade gliomas. Cancer Cell 2020;37:569–83 CrossRef Medline We hypothesized that ADC histogram metrics would be associ- 17. Jones DT, Kocialkowski S, Liu L, et al. Oncogenic RAF1 rearrange- ated with response in pLGGs treated with selumetinib. ADC val- ment and a novel BRAF mutation as alternatives to KIAA1549: ues in pLGGs decreased with time in treatment with selumetinib BRAF fusion in activating the MAPK pathway in pilocytic astrocy- in responders compared with nonresponders. Compared with toma. Oncogene 2009;28:2119–23 CrossRef Medline NF1-associated OPHG, ADC decreased with selumetinib treat- 18. Lassaletta A, Zapotocky M, Mistry M, et al. Therapeutic and prog- nostic implications of BRAF V600E in pediatric low-grade glio- ment in sporadic OPHG. ADC histogram metrics were found to mas. JClin Oncol 2017;35:2934–41 CrossRef Medline be associated with response of pLGG to selumetinib. 19. Karajannis MA, Legault G, Fisher MJ, et al. Phase II study of sorafe- nib in children with recurrent or progressive low-grade astrocyto- Disclosure forms provided by the authors are available with the full text and mas. Neuro Oncol 2014;16:1408–16 CrossRef Medline PDF of this article at www.ajnr.org. 20. Sievert AJ, Lang SS, Boucher KL, et al. Paradoxical activation and RAF inhibitor resistance of BRAF protein kinase fusions charac- terizing pediatric astrocytomas. Proc Natl Acad Sci U S A REFERENCES 2013;110:5957–62 CrossRef Medline 1. Ostrom QT, Cioffi G, Gittleman H, et al. CBTRUS Statistical 21. Burkhard C, Di Patre PL, Schüler D, et al. A population-based study Report: primary brain and other central nervous system tumors of the incidence and survival rates in patients with pilocytic astro- diagnosed in the United States in 2012-2016. Neuro Oncol 2019;21: cytoma. J Neurosurg 2003;98:1170–74 CrossRef Medline v1–100 CrossRef Medline 22. Collins VP, Jones DT, Giannini C. Pilocytic astrocytoma: pathology, 2. Ryall S, Tabori U, Hawkins C. Pediatric low-grade glioma in the era molecular mechanisms and markers. Acta Neuropathol 2015;129:775– of molecular diagnostics. Acta Neuropathol Commun 2020;8:30 78 CrossRef Medline CrossRef Medline 23. Listernick R, Louis DN, Packer RJ, et al. Optic pathway gliomas in 3. Collins KL, Pollack IF. Pediatric low-grade gliomas. Cancers children with neurofibromatosis 1: consensus statement from the 2020;12:1152 CrossRef NF1 Optic Pathway Glioma Task Force. Ann Neurol 1997;41:143– 4. Broniscer A, Baker SJ, West AN, et al. Clinical and molecular char- 49 CrossRef Medline acteristics of malignant transformation of low-grade glioma in 24. Gauvain KM, McKinstry RC, Mukherjee P, et al. Evaluating pediat- children. JClin Oncol 2007;25:682–89 CrossRef Medline ric brain tumor cellularity with diffusion-tensor imaging. AJR Am 5. Sharif S, Ferner R, Birch JM, et al. Second primary tumors in neuro- JRoentgenol 2001;177:449–54 CrossRef Medline fibromatosis 1 patients treated for optic glioma: substantial risks 25. Koral K, Mathis D, Gimi B, et al. Common pediatric cerebellar after radiotherapy. JClin Oncol 2006;24:2570–75 CrossRef Medline tumors: correlation between cell densities and apparent diffusion 6. Jones DTW, Kieran MW, Bouffet E, et al. Pediatric low-grade glio- coefficient metrics. Radiology 2013;268:532–37 CrossRef Medline mas: next biologically driven steps. Neuro Oncol 2018;20:160–73 26. Jost SC, Ackerman JW, Garbow JR, et al. Diffusion-weighted and CrossRef Medline dynamic contrast-enhanced imaging as markers of clinical behav- 7. Banerjee A, Jakacki RI, Onar-Thomas A, et al. A phase I trial of the ior in children with optic pathway glioma. Pediatr Radiol MEK inhibitor selumetinib (AZD6244) in pediatric patients with 2008;38:1293–99 CrossRef Medline recurrent or refractory low-grade glioma: a Pediatric Brain Tumor 27. Yeom KW, Lober RM, Andre JB, et al. Prognostic role for diffusion- Consortium (PBTC) study. Neuro Oncol 2017;19:1135–44 CrossRef weighted imaging of pediatric optic pathway glioma. J Neurooncol Medline 2013;113:479–83 CrossRef Medline 460 Vajapeyam Mar 2022 www.ajnr.org 28. Hsu CH, Lober RM, Li MD, et al. Decreased tumor apparent diffu- 34. Chenevert TL, Stegman LD, Taylor JM, et al. Diffusion magnetic res- sion coefficient correlates with objective response of pediatric low- onance imaging: an early surrogate marker of therapeutic efficacy grade glioma to bevacizumab. JNeurooncol 2015;122:491–96 in brain tumors. J Natl Cancer Inst 2000;92:2029–36 CrossRef CrossRef Medline Medline 29. Payabvash S, Aboian M, Tihan T, et al. Machine learning decision 35. Hoyt WF, Baghdassarian SA. Optic glioma of childhood: natural tree models for differentiation of posterior fossa tumors using dif- history and rationale for conservative management. Br J fusion histogram analysis and structural MRI findings. Front Ophthalmol 1969;53:793–98 CrossRef Medline Oncol 2020;10:71 CrossRef Medline 36. Gross AM, Wolters PL, Dombi E, et al. Selumetinib in children with 30. Vajapeyam S, Brown D, Johnston PR, et al. Multiparametric analysis inoperable plexiform neurofibromas. N Engl J Med 2020;382:1430– of permeability and ADC histogram metrics for classification of 42 CrossRef Medline pediatric brain tumors by tumor grade. AJNR Am J Neuroradiol 37. Lévy P, Bièche I, Leroy K, et al. Molecular profiles of neurofibroma- 2018;39:552–57 CrossRef Medline tosis type 1-associated plexiform neurofibromas: identification of 31. Jaimes C, Vajapeyam S, Brown D, et al. MR imaging correlates for a gene expression signature of poor prognosis. Clin Cancer Res molecular and mutational analyses in children with diffuse intrinsic 2004;10:3763–71 CrossRef Medline pontine glioma. AJNR Am J Neuroradiol 2020;41:874–81 CrossRef 38. Deliganis AV, Geyer JR, Berger MS. Prognostic significance of type Medline 1 neurofibromatosis (von Recklinghausen Disease) in childhood 32. Vajapeyam S, Brown D, Billups C, et al. Advanced ADC histogram, optic glioma. Neurosurgery 1996;38:1114–18; discussion 1118–19 perfusion, and permeability metrics show an association with survival CrossRef Medline and pseudoprogression in newly diagnosed diffuse intrinsic pontine 39. Kornreich L, Blaser S, Schwarz M, et al. Optic pathway glioma: cor- glioma: a report from the Pediatric Brain Tumor Consortium. AJNR relation of imaging findings with the presence of neurofibromato- Am J Neuroradiol 2020;41:718–24 CrossRef Medline sis. AJNR Am J Neuroradiol 2001;22:1963–69 Medline 33. Sugahara T, Korogi Y, Kochi M, et al. Usefulness of diffusion- 40. Wagner MW, Hainc N, Khalvati F, et al. Radiomics of pediatric low- weighted MRI with echo-planar technique in the evaluation of cel- grade gliomas: toward a pretherapeutic differentiation of BRAF- lularity in gliomas. J Magn Reson Imaging 1999;9:53–60 CrossRef mutated and BRAF-fused tumors. AJNR Am J Neuroradiol Medline 2021;42:759–65 CrossRef Medline AJNR Am J Neuroradiol 43:455–61 Mar 2022 www.ajnr.org 461 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png American Journal of Neuroradiology American Journal of Neuroradiology

ADC Histogram Analysis of Pediatric Low-Grade Glioma Treated with Selumetinib: A Report from the Pediatric Brain Tumor Consortium

Download PDF
7 pages

Loading next page...
 
/lp/american-journal-of-neuroradiology/adc-histogram-analysis-of-pediatric-low-grade-glioma-treated-with-RVfPO0UQni

References (42)

Publisher
American Journal of Neuroradiology
Copyright
© 2022 by American Journal of Neuroradiology. Indicates open access to non-subscribers at www.ajnr.org
ISSN
0195-6108
eISSN
1936-959X
DOI
10.3174/ajnr.a7433
Publisher site
See Article on Publisher Site

Abstract

ORIGINAL RESEARCH PEDIATRICS ADC Histogram Analysis of Pediatric Low-Grade Glioma Treated with Selumetinib: A Report from the Pediatric Brain Tumor Consortium S. Vajapeyam, D. Brown, A. Ziaei, S. Wu, G. Vezina, J.S. Stern, A. Panigrahy, Z. Patay, B. Tamrazi, J.Y. Jones, S.S. Haque, D.S. Enterline, S. Cha, B.V. Jones, K.W. Yeom, A. Onar-Thomas, I.J. Dunkel, M. Fouladi, J.R. Fangusaro, and T.Y. Poussaint ABSTRACT BACKGROUND AND PURPOSE: Selumetinib is a promising MAP (mitogen-activated protein) kinase (MEK) 1/2 inhibitor treatment for pediatric low-grade gliomas. We hypothesized that MR imaging–derived ADC histogram metrics would be associated with survival and response to treatment with selumetinib. MATERIALS AND METHODS: Children with recurrent, refractory, or progressive pediatric low-grade gliomas who had World Health Organization grade I pilocytic astrocytoma with KIAA1549–BRAF fusion or the BRAF V600E mutation (stratum 1), neurofibromatosis type 1–associated pediatric low-grade gliomas (stratum 3), or sporadic non-neurofibromatosis type 1 optic pathway and hypothala- mic glioma (OPHG) (stratum 4) were treated with selumetinib for up to 2 years. Quantitative ADC histogram metrics were analyzed for total and enhancing tumor volumes at baseline and during treatment. RESULTS: Each stratum comprised 25 patients. Stratum 1 responders showed lower values of SD of baseline ADC_total as well as a larger decrease with time on treatment in ADC_total mean, mode, and median compared with nonresponders. Stratum 3 responders showed a greater longitudinal decrease in ADC_total. In stratum 4, higher baseline ADC_total skewness and kurtosis were associated with shorter pro- gression-free survival. When all 3 strata were combined, responders showed a greater decrease with time in ADC_total mode and median. Compared with sporadic OPHG, neurofibromatosis type 1–associated OPHG had lower values of ADC_total mean, mode, and median as well as ADC_enhancement mean and median and higher values of ADC_total skewness and kurtosis at baseline. The longitudinal decrease in ADC_total median during treatment was significantly greater in sporadicOPHG comparedwithneurofibromatosis type 1–associated OPHG. CONCLUSIONS: ADC histogram metrics are associated with progression-free survival and response to treatment with selumetinib in pediatric low-grade gliomas. ABBREVIATIONS: MEK ¼ MAP (mitogen-activated protein) kinase; NF1 ¼ neurofibromatosis type 1; OPHG ¼ optic pathway and hypothalamic glioma; PBTC ¼ Pediatric Brain Tumor Consortium; PFS ¼ progression-free survival; pLGG ¼ pediatric low-grade glioma; WHO ¼ World Health Organization ediatric low-grade gliomas (pLGGs) are the most commonly (WHO) grade I pilocytic astrocytoma is the most frequent pri- occurring childhood brain tumor and comprise 40%–50% mary brain tumor in individuals 0–19 years of age, accounting for 1,2 of all childhood CNS tumors. World Health Organization 15% of children’s and adolescents’ (0–19 years) and 17.8% of Department of Radiology (K.W.Y.), Stanford University School of Medicine, Received August 10, 2021; accepted after revision January 1, 2022. Stanford, California; and Department of Hematology, Oncology, and Stem Cell From the Department of Radiology (S.V., T.Y.P.), Boston Children’sHospital, Harvard Transplantation (J.R.F.), Children’s Healthcare of Atlanta and Emory University, Medical School, Boston, Massachusetts; Department of Radiology (D.B.), Massachusetts Atlanta, Georgia. General Hospital, Boston, Massachusetts; Department of Radiology (A.Z.), Boston This work was supported by the American Lebanese Syrian Associated Charities, Children’s Hospital, Boston, Massachusetts; Department of Biostatistics (S.W., A.O.-T.), AstraZeneca, US Department of Health and Human Services, National Institutes of St Jude Children’s Research Hospital, Memphis, Tennessee; Department of Radiology Health, National Cancer Institute, National Cancer Institute Cancer Therapy (G.V.), Children's National Medical Center, Washington, DC; Department of Radiology Evaluation Program, National Institutes of Health/National Cancer Institute Cancer (J.S.S.), Ann and Robert H Lurie Children's Hospital of Chicago, Chicago, Illinois; Center Support Grant P30 CA008748, UM1CA081457. Department of Radiology (A.P.), Children's Hospital of Pittsburgh, Pittsburgh, Pennsylvania; Department of Diagnostic Imaging (Z.P.), St Jude Children's Research Please address correspondence to Sridhar Vajapeyam, MD, Children’s Hospital Hospital,Memphis,Tennessee;Department of Radiology (B.T.), Children's Hospital Los Boston, Radiology, 300 Longwood Ave, Boston, MA 02115; e-mail: Angeles, Los Angeles, California; Department of Radiology (J.Y.J., M.F.), Nationwide sridhar.vajapeyam@childrens.harvard.edu Children's Hospital, Columbus, Ohio; Department of Radiology (S.S.H., I.J.D.), Memorial Indicates open access to non-subscribers at www.ajnr.org Sloan Kettering Cancer Center, New York, New York; Department of Radiology (D.S.E.), Duke University School of Medicine, Durham, North Carolina; Department of Radiology Indicates article with online supplemental data. (S.C.), University of California San Francisco, San Francisco, California; Department of Radiology (B.V.J.), Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio; http://dx.doi.org/10.3174/ajnr.A7433 AJNR Am J Neuroradiol 43:455–61 Mar 2022 www.ajnr.org 455 1 primary childhood (0–14 years) brain tumors. pLGGs are bio- Imaging and Image Analysis logically distinct from their adult counterparts. Unlike adult Standard MR Imaging Evaluation. Standard MR imaging, which low-grade gliomas that occur mostly in the cerebral hemispheres included T2-FLAIR and axial pre- and postcontrast T1-weighted and transform into higher-grade gliomas, pLGGs can occur images, as well as DTI, was performed primarily on 3T scanners throughout the central nervous system, are molecularly distinct, at the start of treatment followed by every 2 months during the 3,4 rarely undergo malignant transformation, and require different first year of therapy and then every 3 months thereafter. therapies. Additionally, high-resolution MR imaging of the orbits and optic Total surgical resection is the first-line of treatment for pLGGs pathway was performed for optic pathway tumors. and can be curative. However, this is not always feasible due to tu- Patients whose tumors achieved an MR imaging response mor location, particularly in the brain stem or optic pathway and (complete or partial response) assessed locally underwent central hypothalamic gliomas (OPHGs), and other therapies are required. radiographic review at the PBTC Neuroimaging Center. A com- Radiation therapy can be effective but may cause considerable neu- plete response was defined as complete tumor disappearance on rocognitive deficits in young children and may increase the risk of T2-FLAIR images, no new lesions, and disappearance of all developing a secondary malignancy, particularly in children with enhancement on T1 postcontrast imaging. A partial response was neurofibromatosis type 1 (NF1). Several chemotherapy regimens defined as at least 50% tumor reduction (in a 2D area calculated as have shown promise in pLGGs, with 5-year progression-free sur- a product of 2 perpendicular linear measurements) on T2-FLAIR. vival (PFS) and overall survival rates of 35%–45% and 85%–100%, Stable disease was defined as neither a sufficient increase nor a respectively, in sporadic pLGG and 5-year 60%–70% PFS and reduction to qualify as a partial response or progressive disease. 90%–100% overall survival rates in NF1-associated pLGG. Progressive disease was a .25% increase or the development of Selumetinib is a potent orally available MAP (mitogen-activated new lesions. These response criteria, while similar to the latest protein) kinase (MEK) 1/2 inhibitor that has recently shown great Response Assessment in Pediatric Neuro-Oncology recommenda- 7,8 efficacy in refractory, recurrent, and progressive pLGGs. We tions, were developed earlier by a PBTC consensus panel and explored the associations of advanced diffusion MR imaging met- used both for local and central imaging response assessments. rics with response and survival in pLGGs treated with selumetinib DTI Acquisition and ADC Histogram Analysis. DTI data were in a large prospective Phase II Pediatric Brain Tumor Consortium acquired with the following acquisition parameters on a 3T scanner: (PBTC) trial, PBTC029B. section thickness = 2.2 mm, TR = 8800 ms, TE = 88 ms, FOV = 220 mm, b-value = 1000 s/mm ,35directions. By meansof the mu- MATERIALS AND METHODS tual information algorithm in FSL (http://www.fmrib.ox.ac.uk/ Subjects fsl), ADC and postcontrast T1 volumes were registered to FLAIR This was a multicenter, National Cancer Institute–sponsored volume as described previously. 3D ROIs comprising the total tu- Phase II study conducted by the PBTC using the MEK 1/2 inhibi- mor volume from the FLAIR images and the enhancing tumor vol- tor selumetinib in patients with pediatric low-grade gliomas ume from postcontrast T1 were automatically generated using the 8,9 treated at 11 PBTC member hospitals. Patients 3–21 years of thresholding feature in Fiji (https://fiji.sc), and the corresponding age with a Lansky/Karnofsky Performance Status score of .60 ADC was used to generate the ADC_total and ADC_enhancement and the presence of recurrent, refractory, or progressive pediatric volumes, respectively. These volumes were thresholded using a uni- 6 2 low-grade glioma after at least 1 standard therapy, including form range of 600–2600  10 mm /s to automatically exclude chemotherapy and radiation therapy, were eligible for inclusion. cysts, necrosis, and hemorrhage; corresponding ADC_total and Patients were assigned to 6 unique strata according to histology, ADC_enhancement histograms were generated. 8,9 tumor location, NF1 status, and BRAF aberration status. Histogram metrics used for statistical analysis were the mean, 8,9 Clinical analyses of strata 1, 3, and 4 have been completed, and SD, mode, median, skewness, and kurtosis of these histograms at the imaging data were used in this study. Stratum 1 comprised baseline and 6, 12, 18, and 24 months into treatment and also at patients with WHO grade I pilocytic astrocytoma with either progression. In studies with multiple lesions, the primary target KIAA1549-BRAF fusion or the BRAF V600E (Val600Glu) muta- lesion was evaluated for all analyses. tion. Stratum 3 comprised patients with imaging- or biopsy- For each of the imaging metrics, the baseline value and the proved pLGGs and a clinical or genetic diagnosis of NF1.Stratum time-dependent longitudinal change in the metrics during the 4 comprised OPHG pLGGs not associated with NF1. course of treatment were examined for associations with tumor Selumetinib was provided as capsules given orally at a dose of volume, response, and PFS. 25 mg/m twice daily in 28-day courses for up to 26 courses. The primary end point was the proportion of patients with a stratum- Statistical Methods specific objective response (partial response or complete The Wilcoxon rank-sum test was used to test the differences in response), as assessed by the local site and sustained for at least ADC parameters at baseline between 2 groups. Subjects with at 8,9 8 weeks. All responses were confirmed by central review. least 1 follow-up MR imaging after baseline were eligible for The protocol was approved by the Cancer Therapy Evaluation time-dependent longitudinal analyses. Tumor volumes were cor- Program (CTEP) as well as each site’s institutional review board. related within the patients because the measures were longitudi- All patients or legal guardians provided written, informed con- nal. The mixed model was used to investigate the correlation sent when applicable based on institutional guidelines. between tumor volume and ADC metrics, taking intrapatient 456 Vajapeyam Mar 2022 www.ajnr.org significantly associated with response (P=.009, q = 0.04), with responders showing lower values than nonres- ponders. The best response achieved was used in all analyses. There were significant associations of response with longitudinal change across time on treatment in the ADC_total mean (P=.02, q = 0.05), ADC_total mode (P=.02, q = 0.07), and ADC_total me- dian (P=.03, q = 0.1), with responders showing a larger decrease in these pa- rameters across time than nonres- ponders. Five of the 8 responders showed a transient increase in ADC followed by a decrease by 24 months as seen in Fig 1. Of the 21 enhancing tumors in stratum 1, sixteen were eli- gible for longitudinal analysis and 8 of these showed a response to selumeti- nib. None of the histogram metrics of ADC_enhancement were associated with a response in the longitudinal analyses. No group differences were found in ADC histogram metrics between FIG 1. The decrease at 24 months in the mean, median, and mode of ADC_total with treatment in a 6-year-old boy with a WHO grade I glioma with a BRAF aberration who responded to selume- the BRAF fusion and the BRAF muta- tinib. The initial transient increase in these metrics seen here was observed in 5 of the 8 respond- tion groups either at baseline or in the ers in stratum 1 and is possibly linked to increased water mobility due to early cell death, longitudinal analysis. followed by tissue consolidation later in treatment. Tx indicates treatment. Stratum 3 Twenty-five patients (mean age at study entry, 10.2 years; range, 3.5–16.6 years; 15 males and 10 correlation into account. Mixed-effects models were used to esti- females) with any NF1-associated pediatric low-grade glioma mate trends across time in the ADC parameters as well as differen- (WHO grades I and II) were enrolled in stratum 3. Nine patients ces in longitudinal changes between groups. Cox models were used showed sustained partial response, 15 had stable disease, and 1 to investigate the association between PFS and ADC parameters at showed progressive disease. The 2-year PFS in stratum 3 was baseline as well as with time. The latter incorporated time-depend- 96%. These results were reported previously. ent ADC parameters. Q values were calculated within each stratum None of the baseline ADC histogram metrics were associated for PFS and response-based analyses separately to adjust for multi- with response or PFS. Longitudinal change with time on treatment plicity via the false discovery rate if.1 parameter was significant. of ADC_total mode was significantly associated with response A q-value of 10% (0.1) was considered significant. Similarly, q-val- (P = .03, q = 0.07), with responders showing a greater decrease ues were also used for cross-strata comparisons. across time than nonresponders. Five of 10 responders showed a transient increase in the ADC followed by a decrease by 24 months. RESULTS Stratum 1 Stratum 4 Twenty-five patients (mean age at study entry, 9.2 years; range, Twenty-five patients (mean age at study entry, 9.4 years; range, 3.9–20.8 years;12males and 13females) with WHO grade I pilo- 3.7–17.6 years; 12 males and 13 females) with sporadic non-NF1 cytic astrocytoma with either 1 of the 2 most common BRAF aber- pediatric optic pathway and hypothalamic low-grade gliomas rations (KIAA1549–BRAF fusion or the BRAF V600E [Val600Glu] (WHO grades I and II) were enrolled in stratum 4. Six patients mutation) were enrolled. Eighteen of these had a KIAA1549–BRAF showed sustained partial response, 14 had stable disease, and 5 fusion, and the remaining 7 had the BRAFV 600E mutation. Nine showed progressive disease. The 2-year PFS in stratum 4 was patients showed sustained partial response, 9 had stable disease, 73.8% (SD 9.3%) as reported previously. and 7 showed progressive disease. Two-year PFS in stratum 1 was At baseline, there were statistically significant associations of 70%. These results were reported previously. ADC_total skewness (P=.02, q = 0.06) and ADC_total kurtosis None of the baseline ADC_total histogram parameters were (P = .02, q = 0.06) with PFS in stratum 4. Patients in stratum 4 associated with PFS. Only the SD of ADC_total at baseline was with higher baseline skewness or kurtosis of ADC_total had AJNR Am J Neuroradiol 43:455–61 Mar 2022 www.ajnr.org 457 OPHGs from stratum 3. No difference was found in PFS between the 2 groups. Enhancing tumor volume and cyst volume at baseline were both significantly lower in the NF1-associ- ated OPHG in stratum 3 compared with stratum 4 (P, .001), while there was no difference in FLAIR tumor vol- ume between the groups. Significant differences between the groups were found in baseline values of ADC_total mean (P = .005, q = 0.01), mode (P = .005, q = 0.01), median (P = .002, q = 0.003), skew- ness (P, .001, q, 0.001), and kurto- sis (P = .01, q = 0.02). The sporadic stratum 4 cohort had higher values of ADC_total mean, mode, and median and lower values of skewness and kur- FIG 2. The longitudinal trend of median ADC of the total tumor volumes for the cohort of 25 tosis. Similarly, ADC_enhancement sporadic non-NF1 OPHGs from stratum 4 shown in red, compared with the 15 NF1-associated showed higher values in sporadic OPHGs shown in blue. Mixed-effects models were used to estimate trends with time in ADC pa- OPHG for mean (P = .045, q = 0.09) rameters as well as differences in longitudinal changes between groups. Sporadic OPHG values and median (P = .04, q = 0.09) com- are higher at baseline (P=.002, q = 0.003) and show a greater decrease with time (P=.02, q =0.06) compared with NF1-associated OPHG. pared with NF1-associated OPHG. In the longitudinal analysis of change with time on treatment, the ADC_total shorter PFS (Online Supplemental Data). None of the longitudi- median decreased significantly more in the sporadic cohort in stra- nal ADC_total histogram metrics were associated with response tum 4 compared with NF1-associated stratum 3 OPHG (P = .02, or PFS. Of the 23 enhancing tumors in stratum 4, twenty were eli- q = 0.06) as shown in Fig 2, with the ADC_total mean similarly gible for longitudinal analysis with only 4 responders in this approaching significance (P = .04, q = 0.11). group. While the entire cohort showed a decrease with time in ADC_enhancement mean (P=.02, q=0.05), mode (P=.01, DISCUSSION q = 0.04), and median (P=.008, q = 0.02), there was no difference The past decade has seen an explosion of molecular data showing between responders and nonresponders. that most pLGGs upregulate the RAS–mitogen-activated protein 15,16 kinase pathway, most commonly including BRAF fusion or Combined Strata 17,18 16 mutation of the BRAF gene and NF1 mutation. More than None of the baseline ADC histogram metrics were associated with 80% of pilocytic astrocytomas have gene alterations in some com- response or PFS when all strata were combined (n = 75). Response ponent of the MEK signaling pathway; therefore, several thera- was still significantly associated with longitudinal change with time pies have been developed to target this pathway. BRAF resistance in treatment for ADC_total mode (P=.02, q = 0.06) and and tumor progression have been reported in some therapies ADC_total median (P = .03, q = 0.09). A greater decrease in these 19,20 such as sorafenib targeting non-V600e aberrations, leading to metrics was found in patients showing a response to the study therapies targeting downstream pathway components like MEK. drug. Of the 64 enhancing tumors in the combined strata, 48 were Selumetinib is one such potent orally available MEK 1/2 inhibitor eligible for longitudinal analysis, with 19 responders. The entire 7-9 that has shown promise in pLGG. cohort showed a decrease with time for the mean (P = .01, Pilocytic astrocytoma is the most frequent primary brain tu- q = 0.04), mode (P = .009, q = 0.03), and median (P = .009, q = 0.03) mor in children and can occur anywhere in the central nervous of ADC_enhancement in the longitudinal analysis, but there was system, with the most common locations being the cerebellum no difference between responders and nonresponders. There were (40%), followed by supratentorial locations (35%), the optic path- no group differences in ADC histogram metrics of tumors that way and hypothalamus (11%), and the brain stem (9%). progressed and those that showed no progression in the 24-month Histopathologically, pilocytic astrocytoma is a tumor of low-to- study period. Total tumor volume was found to be negatively asso- moderate cellularity with compact, densely fibrillated areas rich ciated with the SD of ADC_total (P = .008). in Rosenthal fibers, as well as spongy loosely textured areas com- posed of multipolar cells that have varying degrees of mucoid Optic Pathway and Hypothalamic Glioma: NF1 versus background material, often with microcysts. Most OPHGs, Sporadic Twenty-five subjects with sporadic OPHGs in stratum 4 were including those associated with NF1, are pilocytic astrocytomas. compared with the cohort with 15 NF1-associated low-grade The pathology of these tumors makes it highly likely that the 458 Vajapeyam Mar 2022 www.ajnr.org extracellular matrix is an important contributor to the ADC of associated with shorter PFS, a finding similar to those in previous these tumors. reports in pediatric diffuse intrinsic pontine glioma (DIPG). ADC maps derived from diffusion-weighted images measure Higher skewness and kurtosis signify more homogeneous tumors the diffusivity of water molecules in tissue. In the context of brain with lower ADC and are usually associated with higher cellularity tumors, ADC is influenced by tumor cellularity and the extracel- in high-grade tumors, but in these low-grade OPHGs, they may lular matrix, as well as the presence of edema, cystic components, be indicative of a lower fraction of extracellular matrix. and necrosis. Several studies have reported an inverse relation- Longitudinal ADC_total histogram metrics were not associated ship between ADC and cellularity in a variety of pediatric brain with response or PFS. 24,25 tumors. As cellularity increases, there is less extracellular When data from all 3 strata were combined, there were no space for water diffusion, leading to a lower ADC. Jost et al associations between baseline ADC histogram metrics, and PFS or examined the ADC in a cohort of 14 NF1-associated OPHGs and response, suggesting that strata-specific analyses may be more use- ful. In a study of newly diagnosed OPHGs treated with carboplatin 13 sporadic OPHGs and found that ADC was not associated with 27 27 either NF1 status or clinical aggressiveness. Yeom et al, how- and vincristine, Yeom et al found that tumors with a higher base- ever, found that in a retrospective study of a cohort of 5 patients line mean had a shorter PFS. Our study comprised previously with NF1 and 7 with sporadic pediatric OPHG, a higher baseline treated pLGGs not limited to OPHGs, which may explain why we ADC was predictive of tumor progression and that ADC then found no baseline associations. However, a greater longitudinal declined following subsequent chemotherapy with a standard decrease in mode and median ADC_total with time in treatment combination of carboplatin and vincristine. Similarly, Hsu et al was seen in responders, similar to our findings in the individual reported that the ADC declined during the response to bevacizu- strata. Tumors with smaller total tumor volume were found to mab in a cohort of 8 patients with progressive pLGG. Whereas all have a higher SD of ADC_total, probably due to the higher pro- the above studies used the mean ADC of the tumor volume, we portion of cells on the tumor periphery in smaller tumors. When the longitudinal trends in the individual and combined used ADC histogram analyses to better characterize the ADC dis- strata are considered as a whole, a consistent overall picture tribution in the tumors. ADC histograms derived from the ADC of all voxels in a tu- emerges, suggesting that a steeper drop in ADC_total values dur- mor volume are particularly well-suited to characterize the diffu- ing treatment with selumetinib occurs in patients who respond to sion properties of the entire tumor volume in brain tumors and selumetinib. This finding confirms earlier reports in the literature 27 28 have been used to differentiate pediatric brain tumors by histo- from Yeom et al and Hsu et al, who saw similar trends in 29 30 logic type and tumor grade. ADC histogram parameters have smaller samples of patients with pLGGs undergoing chemother- also been associated with molecular subtype and response to apy. This suggests that monitoring ADC during the course of 12,32 therapy in diffuse intrinsic pontine gliomas in children. treatment may provide some clinical value in assessing the Herein, we report the use of ADC histogram metrics to identify response in pLGGs. A lower ADC has long been shown to be in- prognosis and response criteria in refractory, recurrent, or pro- dicative of higher cellularity, and an increase in ADC has been shown to be a biomarker of cell death in response to treatment in gressive pLGGs treated with selumetinib. Pilocytic astrocytomas in stratum 1 that responded to selume- high-grade gliomas in adults. Our longitudinal results linking a tinib had a smaller SD of baseline ADC_total, suggesting that decrease in ADC with response is not consistent with the oppo- more homogeneous tumors responded better. In addition, the site behavior in high-grade gliomas because these are low-grade mean, median, and mode of ADC_total all decreased more dur- gliomas with a more extensive extracellular matrix. Tumor ing the course of treatment for responders compared with non- response with selumetinib in pLGGs may be the result of a 28 27,35 responders. This decrease is similar to that found by Hsu et al decrease in extracellular space, leading to the decrease in in a small sample of recurrent or progressive pLGGs treated with ADC seen in responders in our study. The impressive response to bevacizumab. The transient increase in these ADC metrics seen selumetinib reported recently in plexiform neurofibromas, in the first few months of treatment is possibly due to increased which are known to have high extracellular matrix content, water mobility due to cell death, followed by tissue consolidation may also be due to changes in the extracellular matrix. Patients with OPHGs associated with NF1 are known to have leading to a stable decrease in these values. Patients in stratum 1 longer PFS compared with those with sporadic OPHGs. When with tumors that had the KIAA1549-BRAF fusion had a longer PFS than those with the BRAF V600E mutation, but there was no we compared the 25 sporadic OPHGs in stratum 4 with 15 NF1- difference in the response rate between the groups. However, associated OPHGs in stratum 3, no significant difference in PFS our analyses of ADC histogram metrics showed no difference was seen, possibly due to the small sample size or the relatively between the BRAF fusion and BRAF mutation either at baseline short period of time of 24 months used in this analysis. or across time, possibly due to small sample size. Enhancing tumor volume as well as the volume of cysts were sig- NF-1 associated pLGG in stratum 3 showed no association nificantly higher in sporadic OPHGs compared with NF1-associ- between baseline ADC histogram metrics and either PFS or over- ated OPHGs, whereas there was no difference in overall tumor all survival. ADC_total mode decreased more with time during volume. The baseline mean, median, and mode of ADC_total as therapy in responders in stratum 3, similar to our findings in stra- well as baseline mean and median of ADC_enhancement were tum 1. higher in sporadic OPHGs compared with NF1-associated Among the patients with sporadic OPHGs in stratum 4, OPHGs. This result may be explained by the higher incidence of higher baseline skewness and kurtosis of ADC_total were cystic components in sporadic OPHGs that we report, consistent AJNR Am J Neuroradiol 43:455–61 Mar 2022 www.ajnr.org 459 39 with previous reports. The greater decrease with time in median 8. Fangusaro J, Onar-Thomas A, Young Poussaint T, et al. Selumetinib in paediatric patients with BRAF-aberrant or neurofibromatosis ADC_total we report in sporadic OPHGs in stratum 4 compared type 1-associated recurrent, refractory, or progressive low-grade with NF1-associated OPHG in stratum 3 (Fig 2) may also be asso- glioma: a multicentre, phase 2 trial. Lancet Oncol 2019;20:1011–22 ciated with a decrease in cystic components of the tumor. CrossRef Medline Although ours was larger than prior studies in the literature, 9. Fangusaro J, Onar-Thomas A, Poussaint TY, et al. Aphase 2trial of the relatively small sample size and diverse prior treatment his- selumetinib in children with recurrent optic pathway and hypo- thalamic low-grade glioma without NF1: a Pediatric Brain Tumor tories are limitations of this study. Consortium Study. Neuro Oncol 2021;23:1777–88 CrossRef Medline Machine learning methods have shown promise in differenti- 10. Fangusaro J, Witt O, Hernáiz Driever P, et al. Response assessment ating BRAF fusion and mutation and could be useful in predict- in paediatric low-grade glioma: recommendations from the ing response to therapy. This study was an observational and Response Assessment in Pediatric Neuro-Oncology (RAPNO) exploratory analysis and is the first study, to our knowledge, to working group. Lancet Oncol 2020;21:e305–16 CrossRef Medline 11. Jenkinson M, Beckmann CF, Behrens TE, et al. FSL. Neuroimage look for any associations between treatment response and 2012;62:782–90 CrossRef Medline changes in ADC in pLGGs. Our population was heterogeneous 12. Poussaint TY, Vajapeyam S, Ricci KI, et al. Apparent diffusion coef- with respect to molecular type as well as tumor location, and ficient histogram metrics correlate with survival in diffuse intrin- response to selumetinib varies by molecular type and tumor loca- sic pontine glioma: a report from the Pediatric Brain Tumor tion. Future studies with the large homogeneous samples of each Consortium. Neuro Oncol 2016;18:725–34 CrossRef Medline 13. Schindelin J, Arganda-Carreras I, Frise E, et al. Fiji: an open-source tumor type required for artificial intelligence and machine learn- platform for biological-image analysis. Nat Methods 2012;9:676–82 ing algorithms may definitively identify ADC histogram metrics CrossRef Medline as markers to identify response to selumetinib. Future work may 14. Strimmer K. fdrtool: a versatile R package for estimating local and also include exploring associations between ADC histogram met- tail area-based false discovery rates. Bioinformatics 2008;24:1461–62 rics and standardized measures of visual outcomes in OPHGs as CrossRef Medline 15. Packer RJ, Pfister S, Bouffet E, et al. Pediatric low-grade gliomas: recommended by Response Assessment in Pediatric Neuro- 10 implications of the biologic era. Neuro Oncol 2017;19:750–61 Oncology. CrossRef Medline 16. Ryall S, Zapotocky M, Fukuoka K, et al. Integrated molecular and CONCLUSIONS clinical analysis of 1,000 pediatric low-grade gliomas. Cancer Cell 2020;37:569–83 CrossRef Medline We hypothesized that ADC histogram metrics would be associ- 17. Jones DT, Kocialkowski S, Liu L, et al. Oncogenic RAF1 rearrange- ated with response in pLGGs treated with selumetinib. ADC val- ment and a novel BRAF mutation as alternatives to KIAA1549: ues in pLGGs decreased with time in treatment with selumetinib BRAF fusion in activating the MAPK pathway in pilocytic astrocy- in responders compared with nonresponders. Compared with toma. Oncogene 2009;28:2119–23 CrossRef Medline NF1-associated OPHG, ADC decreased with selumetinib treat- 18. Lassaletta A, Zapotocky M, Mistry M, et al. Therapeutic and prog- nostic implications of BRAF V600E in pediatric low-grade glio- ment in sporadic OPHG. ADC histogram metrics were found to mas. JClin Oncol 2017;35:2934–41 CrossRef Medline be associated with response of pLGG to selumetinib. 19. Karajannis MA, Legault G, Fisher MJ, et al. Phase II study of sorafe- nib in children with recurrent or progressive low-grade astrocyto- Disclosure forms provided by the authors are available with the full text and mas. Neuro Oncol 2014;16:1408–16 CrossRef Medline PDF of this article at www.ajnr.org. 20. Sievert AJ, Lang SS, Boucher KL, et al. Paradoxical activation and RAF inhibitor resistance of BRAF protein kinase fusions charac- terizing pediatric astrocytomas. Proc Natl Acad Sci U S A REFERENCES 2013;110:5957–62 CrossRef Medline 1. Ostrom QT, Cioffi G, Gittleman H, et al. CBTRUS Statistical 21. Burkhard C, Di Patre PL, Schüler D, et al. A population-based study Report: primary brain and other central nervous system tumors of the incidence and survival rates in patients with pilocytic astro- diagnosed in the United States in 2012-2016. Neuro Oncol 2019;21: cytoma. J Neurosurg 2003;98:1170–74 CrossRef Medline v1–100 CrossRef Medline 22. Collins VP, Jones DT, Giannini C. Pilocytic astrocytoma: pathology, 2. Ryall S, Tabori U, Hawkins C. Pediatric low-grade glioma in the era molecular mechanisms and markers. Acta Neuropathol 2015;129:775– of molecular diagnostics. Acta Neuropathol Commun 2020;8:30 78 CrossRef Medline CrossRef Medline 23. Listernick R, Louis DN, Packer RJ, et al. Optic pathway gliomas in 3. Collins KL, Pollack IF. Pediatric low-grade gliomas. Cancers children with neurofibromatosis 1: consensus statement from the 2020;12:1152 CrossRef NF1 Optic Pathway Glioma Task Force. Ann Neurol 1997;41:143– 4. Broniscer A, Baker SJ, West AN, et al. Clinical and molecular char- 49 CrossRef Medline acteristics of malignant transformation of low-grade glioma in 24. Gauvain KM, McKinstry RC, Mukherjee P, et al. Evaluating pediat- children. JClin Oncol 2007;25:682–89 CrossRef Medline ric brain tumor cellularity with diffusion-tensor imaging. AJR Am 5. Sharif S, Ferner R, Birch JM, et al. Second primary tumors in neuro- JRoentgenol 2001;177:449–54 CrossRef Medline fibromatosis 1 patients treated for optic glioma: substantial risks 25. Koral K, Mathis D, Gimi B, et al. Common pediatric cerebellar after radiotherapy. JClin Oncol 2006;24:2570–75 CrossRef Medline tumors: correlation between cell densities and apparent diffusion 6. Jones DTW, Kieran MW, Bouffet E, et al. Pediatric low-grade glio- coefficient metrics. Radiology 2013;268:532–37 CrossRef Medline mas: next biologically driven steps. Neuro Oncol 2018;20:160–73 26. Jost SC, Ackerman JW, Garbow JR, et al. Diffusion-weighted and CrossRef Medline dynamic contrast-enhanced imaging as markers of clinical behav- 7. Banerjee A, Jakacki RI, Onar-Thomas A, et al. A phase I trial of the ior in children with optic pathway glioma. Pediatr Radiol MEK inhibitor selumetinib (AZD6244) in pediatric patients with 2008;38:1293–99 CrossRef Medline recurrent or refractory low-grade glioma: a Pediatric Brain Tumor 27. Yeom KW, Lober RM, Andre JB, et al. Prognostic role for diffusion- Consortium (PBTC) study. Neuro Oncol 2017;19:1135–44 CrossRef weighted imaging of pediatric optic pathway glioma. J Neurooncol Medline 2013;113:479–83 CrossRef Medline 460 Vajapeyam Mar 2022 www.ajnr.org 28. Hsu CH, Lober RM, Li MD, et al. Decreased tumor apparent diffu- 34. Chenevert TL, Stegman LD, Taylor JM, et al. Diffusion magnetic res- sion coefficient correlates with objective response of pediatric low- onance imaging: an early surrogate marker of therapeutic efficacy grade glioma to bevacizumab. JNeurooncol 2015;122:491–96 in brain tumors. J Natl Cancer Inst 2000;92:2029–36 CrossRef CrossRef Medline Medline 29. Payabvash S, Aboian M, Tihan T, et al. Machine learning decision 35. Hoyt WF, Baghdassarian SA. Optic glioma of childhood: natural tree models for differentiation of posterior fossa tumors using dif- history and rationale for conservative management. Br J fusion histogram analysis and structural MRI findings. Front Ophthalmol 1969;53:793–98 CrossRef Medline Oncol 2020;10:71 CrossRef Medline 36. Gross AM, Wolters PL, Dombi E, et al. Selumetinib in children with 30. Vajapeyam S, Brown D, Johnston PR, et al. Multiparametric analysis inoperable plexiform neurofibromas. N Engl J Med 2020;382:1430– of permeability and ADC histogram metrics for classification of 42 CrossRef Medline pediatric brain tumors by tumor grade. AJNR Am J Neuroradiol 37. Lévy P, Bièche I, Leroy K, et al. Molecular profiles of neurofibroma- 2018;39:552–57 CrossRef Medline tosis type 1-associated plexiform neurofibromas: identification of 31. Jaimes C, Vajapeyam S, Brown D, et al. MR imaging correlates for a gene expression signature of poor prognosis. Clin Cancer Res molecular and mutational analyses in children with diffuse intrinsic 2004;10:3763–71 CrossRef Medline pontine glioma. AJNR Am J Neuroradiol 2020;41:874–81 CrossRef 38. Deliganis AV, Geyer JR, Berger MS. Prognostic significance of type Medline 1 neurofibromatosis (von Recklinghausen Disease) in childhood 32. Vajapeyam S, Brown D, Billups C, et al. Advanced ADC histogram, optic glioma. Neurosurgery 1996;38:1114–18; discussion 1118–19 perfusion, and permeability metrics show an association with survival CrossRef Medline and pseudoprogression in newly diagnosed diffuse intrinsic pontine 39. Kornreich L, Blaser S, Schwarz M, et al. Optic pathway glioma: cor- glioma: a report from the Pediatric Brain Tumor Consortium. AJNR relation of imaging findings with the presence of neurofibromato- Am J Neuroradiol 2020;41:718–24 CrossRef Medline sis. AJNR Am J Neuroradiol 2001;22:1963–69 Medline 33. Sugahara T, Korogi Y, Kochi M, et al. Usefulness of diffusion- 40. Wagner MW, Hainc N, Khalvati F, et al. Radiomics of pediatric low- weighted MRI with echo-planar technique in the evaluation of cel- grade gliomas: toward a pretherapeutic differentiation of BRAF- lularity in gliomas. J Magn Reson Imaging 1999;9:53–60 CrossRef mutated and BRAF-fused tumors. AJNR Am J Neuroradiol Medline 2021;42:759–65 CrossRef Medline AJNR Am J Neuroradiol 43:455–61 Mar 2022 www.ajnr.org 461

Journal

American Journal of Neuroradiology – American Journal of Neuroradiology

Published: Mar 1, 2022

There are no references for this article.