Historically, individuals with neurofibromatosis type 1 (NF1) who have a symptomatic plexiform neurofibroma (PN) were given one option for treatment: partial surgical resection. The advent of effective targeted medical intervention with oral MEK inhibitors that shrink PNs in patients with NF1 paves the way for new approaches in NF1 management. The underpinning of this dramatic change in care for patients is magnetic resonance imaging (MRI); specifically, the implementation of volumetric MRI to assess effectiveness of therapeutic intervention. The article of Cai and Steinberg in this issue compares two methods of volumetric analysis for PNs, which demonstrate that application of standard methods in radiologic review enables practitioners to determine when to initiate treatment and evaluate effectiveness of medical intervention.
PNs are benign heterogeneous cell collections of Schwann cells, fibroblasts, mast cells, and neovascular cellular components that infiltrate nerve sheaths and envelop nerve bundles. Unlike dermal neurofibromas that affect almost all adults with NF1, PNs are seen in a minority (20%) of patients with NF1. There are multiple types of PNs, but all originate from clonal expansion of Schwann cells that have undergone double inactivation of both NF1 alleles, leading to paracrine recruitment of multiple cell types . The NF1 tumor-suppressor gene encodes neurofibromin, a Ras-GTPase activating protein that decreases activated-ras intracellular signaling through the MAPK pathway. Loss of intracellular neurofibromin leads to higher levels of activated wild-type ras with heightened propagation of downstream signaling through Raf, MEK, and Erk, eventually leading to increased cell proliferation and survivability. As slow-growing benign tumors, PNs are unresponsive to untargeted chemotherapy protocols or radiation regimens. Surgical resection for symptomatic PNs results in neurologic deficits because of the sacrifice of intertwined nerves within the tumors. The genome of benign PNs is stable, and only a small percentage of tumors transform to sarcomas, usually malignant peripheral nerve sheath tumors, after accumulation of multiple genetic changes in other cancer-related pathways, such as TP53. Transformed PNs and atypical neurofibromatous neoplasms of uncertain biologic potential will likely render a tumor unresponsive to MEK inhibitors that target the Ras-MAPK pathway.
PNs can arise from any peripheral nerve sheath. Isolated nodular PNs tend to be “corralled” within the nerve sheaths and remain localized to the area from which it emerges. Symptomatic nodular PNs can be successfully resected, although, even with nerve-sparing surgical approaches, there is residual loss of nerve function. On the other hand, diffuse PNs tend to track along peripheral nerve sheaths, leading to body asymmetry and may affect entire limbs (see Fig 1, the article of Cai and Steinberg et al., in this issue, PN 15). Surgical resection of such diffuse and invasive tumors is impractical and usually attempted only if there is concern for malignant transformation to malignant peripheral nerve sheath tumor. Paraspinal PNs arise from nerve roots and may coalesce along multiple foramina (see Fig 1, PN 5 along the upper thoracic spine). As depicted in the MRI images in the article by Cai and Steinberg, patients often have multiple PNs, which could occur either by multiple independent somatic inactivation of the normal NF1 allele (second hits) or by early embryonic seeding of neural crest-derived progenitor Schwann cells that undergo singular somatic inactivation of the normal NF1 allele followed by migratory dissemination. Symptoms of PNs are highly variable and dependent on size and location of the tumors. A primary symptom is neurogenic pain and secondary symptoms are because of interference of normal tissue function by compression of adjacent tissue. Progression of PNs is unpredictable, and some remain dormant after they emerge, usually in early to middle childhood. In general, asymptomatic PNs are monitored by clinical evaluation and intermittent MRI studies. For symptomatic PNs, the location, size, and degree of nerve involvement determine symptoms, and documentation by various MRI modalities enable surgeons and oncologists to make decisions on intervention.
MRI has been the mainstay of neurogenic tumor imaging since its incorporation into clinic practice a few decades ago. Short tau inversion recovery and standard T2-weighted sequences with fat saturation are optimal to visualize PNs. Metabolically active components of PNs are detected by both contrast enhancement and positron emission tomography-computed tomography imaging . Regional MRI with gadolinium enhancement provides the highest resolution to detect and characterize symptomatic PNs. In asymptomatic individuals with NF1, “whole-body MRI” has been implemented to determine a patient’s tumor load, usually sometime after puberty when the likelihood of emergence of new PNs subsides . Whole-body MRI for all individuals with NF1 has the advantage of one-stop assessment of tumor load, and once tumors have been identified and symptoms were ascertained, a more focused MRI surveillance plan can be implemented for long-term management of PNs. Individualized management plans for assessment of PN progression, malignant transformation, or response to therapy require the implementation of a quantitative assessment of tumor size.
The purpose of the Cai and Steinberg study is to compare the level of agreement between two separate semi-automated methods of volumetric MRI analyses for PNs; one developed at the National Cancer Institute (NCI) Center for Cancer Research, Pediatric Branch (E Dombi) called MEDx and one developed at Massachusetts General Hospital (MGH), 3D Imaging Service (G Harris) called 3DQI. The MEDx software uses a slice-by-slice histogram analysis of selected areas to identify the signal intensity threshold between tumor and normal tissue, and requires up to 1 hour per patient examination. The 3DQI method generates a 3D rendering of the image data with the tumor surface identified by the dynamic-threshold level set method starting with a seed initiation within the lesion and propagating shell expanding to the boundary, and requires approximately 10–20 minutes per patient examination. Using MRI data from the same 15 PNs (13 of which were progressive) at multiple sequential time points, the two techniques, 3DQI and MEDx, were compared in three ways; MGH-3DQI versus NCI-3DQI (same technique analyzed at two separate centers), NCI-3DQI versus NCI-MEDx (two techniques analyzed at one center), and MGH-3DQI versus NCI-MEDx (two techniques analyzed at the respective center that developed the technique). The key question is whether either technique can be used in volume assessment. The volume differences were less than 20% in 117 of 135 volume-pair comparisons, and differences in interval volume change were less than 20% in 120 of 135 comparisons. The authors conclude that this study demonstrates good agreement in percent volume changes between time points and tumor progression status classification. They recognize that selection of either validated technique will depend on reliability, ease of use, and processing speed for adaptation to clinical practice. Within the clinical research setting, either of these relatively labor-intensive techniques would be feasible. However, widespread acceptance and utilization of volumetric imaging in clinical radiology will require the seamless integration of such software into the Picture Archiving and Communication System environment. Such projects are already underway with many imaging technology companies building 3D volumetric segmentation into their software packages. As they currently exist, such packages require manual input from the radiologist and remain somewhat inefficient. The final iteration will be fully automated tumor segmentation using machine learning to calculate 3D tumor volumes with minimal human input . Studies such as this from Cai and Steinberg are important first steps in validating these segmentation techniques and will form the foundation for more automated approaches to come.
The significance of these results for patients with NF1 rests with the identification of a successful medication that targets the Ras-MAPK signal transduction pathway in PNs. The MEK inhibitors will likely be approved by the Food and Drug Administration for treatment of PNs, and with this option for care, academic centers affiliated with NF Clinics will have a significant number of patients needing more quantitative MRI monitoring. Accurate quantitation of PN volume will be instrumental in the determination of which PNs may benefit from initiation of therapy. Quantification of tumor volume will be essential to determine response to therapy, as almost one-third of NF1-related PNs do not respond to MEK inhibitor .
Get Radiology Tree app to read full this article<
References
1. Dombi E., Baldwin A., Marcus L.J., et. al.: Activity of selumetinib in neurofibromatosis type 1-related plexiform neurofibromas. N Engl J Med 2016; 375: pp. 2550-2560.
2. Straser K., Yang F.-C., Clapp D.W.: Pathogenesis of plexiform neurofibroma: tumor-stromal/hematopoietic interactions in tumor progression. Annu Rev Pathol 2012; 7: pp. 469-495.
3. Miettinen M., Antonescu C., Fletcher C., et. al.: Histopathologic evaluation of atypical neurofibromatous tumors and their transformation into malignant peripheral nerve sheath tumor in patients with neurofibromatosis 1: a consensus overview. Hum Pathol 2017; 67: pp. 1-10.
4. Benz M., Czernin J., Dry S., et. al.: Quantitative F18-fluorodeoxyglucose positron emission tomography accurately characterizes peripheral nerve sheath tumors as malignant or benign. Cancer 2010; 116: pp. 451-458.
5. Ahlawat S., Fayad L., Khan M.S., et. al.: Current whole-body MRI applications in the neurofibromatoses: NF1, NF2, schwannomatosis. Neurology 2016; 87: pp. S31-S39.
6. Plotkin S., Bredella M., Cai W., et. al.: Quantitative assessment of whole-body tumor burden in adult patients with neurofibromatosis. PLoS ONE 2012; 7: e35711
7. Lakhani P., Prater A.B., Hutson R.K., et. al.: Machine learning in radiology: applications beyond image interpretation. J Am Coll Radiol 2017; [Epub ahead of print]