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

Learn More →

How Effective Are Noninvasive Tests for Diagnosing Malignant Peripheral Nerve Sheath Tumors in Patients with Neurofibromatosis Type 1? Diagnosing MPNST in NF1 Patients

How Effective Are Noninvasive Tests for Diagnosing Malignant Peripheral Nerve Sheath Tumors in... Hindawi Sarcoma Volume 2019, Article ID 4627521, 8 pages https://doi.org/10.1155/2019/4627521 Research Article How Effective Are Noninvasive Tests for Diagnosing Malignant Peripheral Nerve Sheath Tumors in Patients with Neurofibromatosis Type 1? Diagnosing MPNST in NF1 Patients Maria Schwabe , Stanislav Spiridonov, Elizabeth L. Yanik, Jack W. Jennings, Travis Hillen, Maria Ponisio, Douglas J. McDonald, Farrokh Dehdashti, and Cara A. Cipriano Washington University School of Medicine in St. Louis, 660 South Euclid Avenue, St. Louis, MO 63110, USA Correspondence should be addressed to Cara A. Cipriano; cipriano@wustl.edu Received 20 February 2019; Accepted 30 May 2019; Published 1 July 2019 Academic Editor: Manish Agarwal Copyright©2019MariaSchwabeetal.&isisanopenaccessarticledistributedundertheCreativeCommonsAttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Background. Distinguishing between benign and malignant peripheral nerve sheath tumors (MPNSTs) in neurofibromatosis 1 (NF1) patients prior to excision can be challenging. How can MPNST be most accurately diagnosed using clinical symptoms, magnetic resonance imaging (MRI) findings (tumor size, depth, and necrosis), positron emission tomography (PET) measures (SUV , SUV , SUV /SUV , and qualitative scale), and combinations of the above? Methods. All NF1 patients peak max max tumor mean liver who underwent PET imaging at our institution (January 1, 2007–December 31, 2016) were included. Medical records were reviewedforclinicalfindings;MRimagesandPETimageswereinterpretedbytwofellowship-trainedmusculoskeletalandnuclear medicine radiologists, respectively. Receiver operating characteristic (ROC) curves were created for each PETmeasurement; the areaunderthecurve(AUC)and thresholdsfordiagnosing malignancywere calculated.Logisticregressiondeterminedsignificant predictors of malignancy. Results. Our population of 41 patients contained 34 benign and 36 malignant tumors. Clinical findings did not reliably predict MPNST. Tumor depth below fascia was highly sensitive; larger tumors were more likely to be malignant but without a useful cutoff for diagnosis. Necrosis on MRI was highly accurate and was the only significant variable in the regression model. PETmeasures were highly accurate, with AUCs comparable and cutoff points consistent with prior studies. A diagnostic algorithm was created using MRI and PET findings. Conclusions. MRI and PET were more effective at diagnosing MPNST than clinical features. We created an algorithm for preoperative evaluation of peripheral nerve sheath tumors in NF1 patients, for which additional validation will be indicated. Differentiating between benign and malignant PNSTs 1. Introduction can be challenging, especially in individuals with multiple Neurofibromatosis type 1 (NF1) is one of the most common neurofibromas. Traditionally, this has been attempted based autosomal-dominant diseases worldwide [1–8]. It has an in- on imaging characteristics and symptoms, which may in- clude pain, increasing size of a mass, and new neurological cidenceof1/2,500to1/3,500individuals[1–5,8,9]andiscaused bymutationsoftheNF1 genelocatedonchromosome17q11.2. deficit [5]. However, there is significant overlap in the ap- Clinically, the disease is characterized by multiple plexiform pearance as well as clinical manifestations of benign and neurofibromas that are usually benign; however, they have the malignant tumors [5, 12]. potential for malignant transformation, with a lifetime risk of Several magnetic resonance imaging (MRI) features can 8–12% [1–5, 8]. Malignant peripheral nerve sheath tumors be useful in distinguishing MPNSTs from neurofibromas. (MPNSTs) are the leading cause of mortality in NF1, reducing &ese include largest dimension of the mass, heterogeneity average life expectancy by 10–15 years [2, 6, 10, 11]. indicating necrosis, peripheral enhancement pattern, 2 Sarcoma perilesional edema-like zone, intratumoral cystic lesion, and on T1 sequences and analyzed as both a continuous and irregular margins [12, 13]. Of these features, tumor size and categorical variable, thelatter using5cmas acutoffbased on necrosis are the best supported predictors for diagnosing the AJCC staging system. Tumor depth was analyzed as a MPNST [12, 14–17] (Figure 1(a)). categorical variable, either superficial or deep to the fascia. Multiple efforts have been made to accurately diagnose Necrosis was defined as nonenhancement on T1 fat- malignant transformation using metabolic imaging with saturated postcontrast images, often with increased T2 positron emission tomography/computed tomography signal intensity, and recorded in quartiles (0%, <25%, 25– (PET/CT)and[ F]fluorodeoxyglucose(FDG)(Figure1(b)). 49%, 50–74%, and ≥75% necrosis). Several authors have studied both semiquantitative and qualitative methods of evaluating lesions with overall good 2.3. FDG-PET/CT Protocol. Patients were imaged according success [8, 18, 19]. Parameters for semiquantitative analysis to our institution’s standard protocol for FDG-PET/CT, include but are not limited to mean standardized uptake which has previously been described [22]. In brief, pa- value(SUV ),maximumSUV(SUV ),maximumSUV mean max tients fasted for at least 4 hours prior to FDG injection. corrected for lean body mass (SUL ), and various ratios max Bloodglucoselevelswererequiredtobe≤200mg/dLpriorto comparing tumor FDG avidity to that of other tissues, such FDG administration. Patients were injected with approxi- as the liver, muscle, and fat. Among the most common of mately 10–15mCi (370–555MBq) of FDG. &e PET/CT these ratios is SUV /SUV , also referred to as max tumor mean liver acquisition was started approximately 60 minutes after the tumor-to-liver ratio (TLR). Semiquantitative methods FDG injection. Patients were scanned from the base of the for diagnosing MPNST, such as mean SUV , have sen- max skull to the upper-thigh with extension to extremities based sitivities of 94%–100% and specificities of 76%–94% [18]. on the location of the lesion of the interest. Noncontrast CT Similarly, qualitative methods, such as visual descriptions of images were obtainedfirst for attenuation correction and for hypermetabolic lesions, have yielded sensitivities of 91%– fusion with the PET images for lesion localization. PET 100% and specificities of 67%–95% [18, 20, 21]. images were acquired at typically 6–8 bed positions, with an In spite of thisresearch, to our knowledge, a noninvasive acquisition time of 2–5 minutes per bed position. gold standard algorithm for diagnosing MPNST has not been established, nor have imaging modalities been evalu- ated in combination with clinical features. &e aim of our 2.4. Semiquantitative Analysis. SUVs were measured using retrospective study was to develop a strategy for dis- the commercial software Hermes (Hermes Medical Solu- tinguishing benign from malignant PNSTusing noninvasive tions, Sweden) by placing a volume of interest (VOI) with a observations and tests, specifically clinical symptoms, MRI diameter of 1.5cm over the most intense region of the lesion features (size, depth, and necrosis), PETmeasures (SUV , on the axial PET/CT images. For SUV , the highest SUV peak max SUV , SUV /SUV , and qualitative scale), and for SUV , the average of the highest SUV within the max max tumor mean liver peak and combinations of the above. VOIs were measured and recorded. When necessary, the diameter of the sphere was adjusted to accommodate for lesion size. Mean liver SUV (average SUV within the VOIs) 2. Materials and Methods was measured using the same 1.5cm diameter sphere placed over the right lobe of the liver. TLR was calculated using 2.1. Patient Population and Data Collection. Following IRB SUV of the tumor over SUV of the liver. approval, all patients with a diagnosis of NF1 who were max mean treated at our institution between 1 January 2007 and 31 December 2016 were identified by searching the medical 2.5. Qualitative Analysis. FDG-PET/CT images were eval- oncology, nuclear medicine, and surgical databases at our uated by one nuclear medicine physician and one nuclear institution. Our study included all patients who underwent medicine fellow who had completed a diagnostic radiology FDG-PET/CT to evaluate for potential malignant trans- residency. Sites of abnormal metabolic activity were scored formation of a PNST with available imaging and confirmed on a 5-point scale based on the following criteria: score of histopathology. Electronic medical records were reviewed for 1 �uptake similar to background, score of 2 �uptake greater demographic information (patient age at time of surgery, than background but less than mediastinal blood pool gender, andtumorlocation)as wellas potential predictors for (MBP), score of 3 �uptake>MBP but less than or equal to malignancy. &ese included preoperative MRI features (tu- liver, score of 4 uptake>liver, and score of 5 �uptake morsize,tumordepthrelativetothefascia,andnecrosis)[22], markedly>than liver (greater than 2-3 times). PETimaging measures (SUV , SUV , and SUV / max peak max tumor SUV ), and clinical findings (pain, enlargement, and mean liver 2.6. Statistical Analysis. Tumor necrosis was analyzed as a nerve symptoms). Histopathology results from biopsy or categorical variable, once with any necrosis considered a surgery obtained through chart review were used as the gold positive finding and once with necrosis >25% considered a standard for diagnosing benign versus malignant PNST. positive finding. For each PET measurement, receiver op- erating characteristic curves (ROC) were created, and the 2.2. MRI Analysis. MR images were assessed by two areaunderthecurve(AUC)wascalculated,alongwithcutoff fellowship-trained musculoskeletal radiologists blinded to points for diagnosing malignancy that optimized sensitivity diagnoses. Tumor size was measured as the largest diameter and specificity. When these were conflicting, sensitivity was Sarcoma 3 5.5 (a) (b) Figure 1: Example of a malignant peripheral nerve sheath tumor in the adductor musculature of the left thigh visualized using (a) magnetic resonance imaging (MRI) and (b) positron emission tomography (PET) studies. &e hypointense areas marked with the asterisk represent necrosis within the tumor. prioritized in order to minimize false negative results in the cases (51.4%). Most patients reported pain and enlargement context of evaluating for malignancy. &ese cutoff points of tumors but not associated nerve symptoms. Additional andtheirsensitivity/specificityinourdatasetwerecompared descriptive statistics including demographic information to cutoff points previously reported in the literature. &e and potential predictors of malignancy are summarized in Table 1. sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated for each tumor characteristic, imaging measure, and clinical finding, 3.1. Clinical Findings. Clinical findings were available in 41 and relevant combinations of these variables. Logistic re- gression was used to evaluate variables for predicting ma- patients for 60 of the 70 tumors. Among patient symptoms, we found that pain and enlargement were more sensitive lignancy in combination. Superficial tumors were not included in the model because no malignancies were ob- (sensitivity 85.2% and 85.2%; specificity 28.1% and 25.0%, respectively), while nerve symptoms were more specific served within this group. Potential predictors (tumor size, necrosis, PET parameters, and clinical findings) were se- (sensitivity 44.8%; specificity 75.0%). Combining nerve symptoms and either pain or growth did not improve di- lected for entry into the model based on the results of the accuracy analysis above. agnostic accuracy compared to evaluation of nerve symp- toms alone. Of note, having any two of three symptoms was more specific and slightly more sensitive than having any 2.7. Diagnostic Algorithm. Our aggregate results were used one particular symptom. Having all three symptoms was to create an algorithm for suggested evaluation of PNST in insensitive and only moderately specific and therefore not NF1 patients. In order to develop a clinically relevant diagnostically useful (Table 2). workup strategy, tumors were first differentiated by tumor features that can be determined by history and physical examination, then by noninvasive imaging studies, and fi- 3.2. MRI Findings. Fifty-nine lesions had PET/CTimages, of nally by biopsy. For each branch point, we calculated the which 37 had MRI images available for review. Tumor depth NPV and PPV to inform clinicians about the likelihood of below the fascia was highly sensitive for malignancy (100%) malignancy given the available information. Lastly, we used with 100% NPV, but not specific (20.7%). Increasing tumor the PPV from the noninvasive workup to offer recom- size was predictive of malignancy as a continuous variable in mendations for management following biopsy, taking pre- regressionanalysis;however,thecutoffof5cm(basedonthe test probability into account. AJCC system) was neither sensitive nor specific. An ROC curve was also constructed for tumor size, but with AUC 0.687 (CI 0.539–0.835), there was no clear threshold for 3. Results diagnosing malignancy. Between the two musculoskeletal radiologists reviewing Our population of 41 patients contained 34 benign and 36 malignanttumors.&e mean patientagewas 30 yearsoldfor MRI studies, interobserver agreement was good (kappa the overall population (range 9–62). Within the 41 patients, 0.608, 95% CI 0.409–0.807) for necrosis by quartiles (none, there was a predominance of tumors in females (41) com- 0–25%, 25–50%, etc.). Agreement was very good for no pared to males (29). &e majority of tumors were deep and necrosis versus any necrosis (kappa 0.937, 95% CI: 0.816– 1.000) and necrosis less than versus greater than 25% (kappa axially located, and the mean diameter was 6.5cm (range 1.5–20.0cm). Mean SUV , SUV , and TLR were 7.9 0.852, 95% CI: 0.655–1.000). For diagnosing MPNST, any max peak necrosis seen on MRI was sensitive (87.5%) and fairly (standard deviation (SD)±5.4), 6.4 (SD±4.3), and 4.0 (SD±2.6), respectively. Necrosis was evident on MRI in 19 specific (76.1%), whereas necrosis of >25% of the tumor SUV 4 Sarcoma Table 1: Descriptive statistics and potential predictors of malig- findings on either MRI (any necrosis) or PET nancy for the 70 peripheral nerve sheath tumors in our study (SUV >4.5) were highly sensitive (95.5%), while posi- peak population of 41 NF1 patients. &ese included patient character- tive findings on both MRI and PET were highly specific istics (age and gender), PET findings (SUV , SUV , and TLR), max peak (93.9%). MRI findings (size, depth relative to the fascia, and necrosis) and Logistic regression analysis supported the value of ne- clinical findings (pain, tumor enlargement, and nerve symptoms), crosis and PETmeasures for diagnosing MPNST. Depth was and histologic diagnosis. found to have 100% sensitivity, with all malignancies oc- Mean 30 curring deep to the fascia in our population; therefore, only Age (years) Median 28 Patient 27 deep tumors were included for the development of Range 9–62 characteristics prediction models. In order to prevent multicollinearity Gender (# of tumors Female 41 between similar measures, SUV >4.5 was selected to each) Male 29 peak represent PET measures, and necrosis >25% of the tumor Yes 46 Pain volume was chosen to represent MRI findings. &us, tumor No 14 size, necrosis >25%, and SUV were selected as the rel- Yes 46 peak Clinical findings Enlargement evant variables; when these were entered for deep tumors, No 14 Yes 23 necrosis was the only significant predictor (p � 0.033, Nerve symptoms No 37 Nagelkerke R �0.622). Mean 6.5 Size (cm) Median 5.4 Range 1.5–20.0 3.4. Combined Diagnostic Algorithm. Clinical findings were MRI findings Depth (relative to Superficial 7 not included in the algorithm as they could not reliably rule fascia) Deep 63 out malignancy in our population (NPV 62.3–69.2%). Yes 19 SUV was included because it had the highest sensitivity Necrosis peak No 18 and NPV of the semiquantitative PETparameters; however, Mean 7.9 the visual scale was almost identical, and the other PET SUV Median 7.4 max measures were not significantly different, so these could be Range 0.7–22.6 substituted for SUV with minimal effect. peak Mean 6.4 Tumors were first differentiated by depth relative to the PET findings SUV Median 5.0 peak fascia, which was notable for 100% NPV in our analysis. Range 0.5–18.7 While tumor depth may be verified on MRI (as in our Mean 4.0 methods), in most cases, it can be easily determined by TLR Median 3.7 Range 0.4–11.9 physical examination prior to advanced imaging. We therefore suggest that patients appearing to have deep tu- Yes 36 Histology Malignant No 34 mors on physical examination be evaluated with MRI and PET and that they undergo biopsy in the presence of con- cerning features on either of these studies. Tumors that are volume was less sensitive (75.0%) but more specific (95.2%), histologically confirmed to be malignant should be managed with PPV 92.3%. with surgery and neoadjuvant/adjuvant chemotherapy/ radiation according to standard protocols. Unfortunately, biopsy itself is imperfect due to sampling error [25]. For 3.3. PET Findings. &e ROC curves for SUV , SUV , max peak tumors that are likely malignant based on imaging but do and TLR were noted to have similar AUCs (Table 3). Op- not contain evidence of MPNST on initial biopsy, rebiopsy timal cutoff points were chosen to maximize sensitivity and or wide excision may be indicated; in contrast, observation specificity based on our patient population. Our cutoff may be acceptable for tumors that appear less concerning points for SUV (4.5) and TLR (3.0) were identical to peak (Figure 2). Of note, this algorithm is based solely on our those in the prior literature, and our cutoff for SUV was max patient population and will therefore require further similar (5.3 in our data compared to 5.0 in the literature, validation. with some more conservative cutoffs slightly lower) [13, 23, 24]. In our dataset, the previously established cutoff of 5.0 resulted in identical sensitivity but lower specificity 4. Discussion (60% compared to 70%) than our optimal cutoff point of 5.3 (Table 3). &e aim of our study was to report the diagnostic value of For qualitative assessment,the interobserver agreement tumor size and depth, MRI features, PET measures, and was verygood, withkappa0.896, 95%CI 0.740–1.000.Level clinical findings to distinguish between benign and malig- 5/5 on the visual scale was considered suggestive of ma- nant PNST. In sum, PET measures and necrosis on MRI lignancy. Using these thresholds for diagnosing MPNST, were the most predictive of malignancy and can be com- PET measures were comparable to one another, with bined to direct workup and treatment in this challenging generallygoodpredictivevalue.Ofthese,SUV >4.5had clinical situation. &is algorithm has been adopted at our peak the highest sensitivity and NPV and was therefore selected institution and will require validation with long-term for testing in combination with MRI necrosis. Positive follow-up from multiple centers. Sarcoma 5 Table2:Sensitivity,specificity,positivepredictivevalues,andnegativepredictivevalues,alongwiththeirassociatedconfidenceintervals,for potential predictors of malignancy. &ese included tumor characteristics (depth relative to the fascia, and size>5cm in diameter), MRI findings (any necrosis, or necrosis>25% by volume), PET findings (SUV >5, SUV peak>4.5, and TLR>3), and clinical findings (pain, max enlarging, and nerve symptoms). Combinations of imaging findings (SUV >or<4.5 on PET and any or >25% necrosis on MRI) and peak clinical findings (1 of 3, 2 of 3, or 3 of 3 symptoms) are also included. Sensitivity Specificity PPV NPV Accuracy 100.0 20.7 61.0 100.0 Deep 64.6 (88.0–100.0) (8.7–40.3) (47.4–73.2) (51.7–100.0) Tumor characteristics 68.0 53.5 56.7 65.2 Size>5cm 60.4 (46.4–84.2) (34.2–72.0) (37.7–74.0) (42.8–82.8) 87.5 76.1 51.4 73.7 Necrosis (any) 81.1 (60.4–97.8) (52.4–90.9) (34.7–67.8) (32.2–65.3) MRI findings 75.0 95.2 92.3 83.3 Necrosis (>25%) 86.5 (47.4–91.7) (74.1–99.8) (62.1–100) (61.8–94.5) 89.3 73.3 75.8 88.0 SUV >5 78.0 max (70.6–97.2) (50.8–87.0) (57.4–88.2) (67.7–96.8) 94.7 70.8 72.0 94.4 SUV >4.5 78.0 peak (71.9–99.7) (48.8–86.6) (50.4–87.1) (70.6–99.7) PET findings 91.7 73.9 78.6 89.5 TLR>3 82.1 (71.5–98.5) (51.3–88.90 (58.5–91.0) (65.5–98.2) 94.1 70.8 69.6 94.4 Visual score 80.5 (69.2–99.7) (48.7–86.6) (47.0–86.0) (70.6–99.7) SUV >4.5 (PET) or any necrosis 95.5 66.7 67.7 95.2 peak 78.8 (MRI) (75.1–100) (47.1–82.1) (48.5–82.7) (74.1–100) SUV >4.5 (PET) and any necrosis 56.2 93.9 81.8 81.6 peak 81.6 Combined PET/MRI (MRI) (30.6–79.2) (78.4–98.9) (47.8–96.8) (65.1–91.7) findings SUVpeak>4.5(PET)ornecrosis>25% 72.2 50.0 60.1 64 (42.8–81.2) 61.1 (MRI) (46.5–90.3) (26.0–74.0) (36.7–78.5) SUVpeak>4.5 (PET) and necrosis 61.5 100 100 78.3 83.9 >25% (MRI) (31.6–86.14) (81.5–100) (59.8–100) (64.4–87.6) 85.2 28.1 50 69.2 Pain 54.2 (65.4–95.1) (14.4–47.0) (35.1–64.9) (38.9–89.6) 85.2 25.0 48.9 66.7 Clinical findings Growth 52.5 (65.4–95.1) (12.1–43.8) (34.3–63.7) (35.4–88.7) 44.8 75.0 59.1 62.3 Nerve symptoms 61.0 (26.9–64.0) (57.5–87.3) (36.7–78.5) (46.7–76.6) 92.9 9.68 48.1 60.0 1 of 3 findings 49.1 (75.0–98.8) (2.53–26.9) (34.5–62.0) (17.0–92.7) Combined clinical 89.3 43.8 58.1 82.3 2 of 3 findings 65.0 findings (70.6–97.2) (26.8–62.1) (42.2–72.6) (55.8–95.3) 32.1 78.1 56.2 43.2 3 of 3 findings 56.7 (16.6–5.24) (60.0–90.1) (30.6–79.2) (28.7–58.9) Our study had several limitations. It is a retrospective Table 3: Areas under the curve (AUC) and associated confidence analysiswitharelativelysmallsamplesizeduetotherarityof intervals (CI) from receiver operating characteristic curves for NF1 and MPNST. Most importantly, PET/CT, MRI, and SUV , SUV , and tumor-to-liver ratio (TLR). Using these max peak clinical findings were not available for all patients; however, present data, cutoff points were selected to optimize sensitivity and these missing data did not correlate with year, age, or specificity. Our cutoff point for SUV differed slightly from that max established in the previous literature, so the sensitivity and spec- malignancy characteristics. In addition, PET/CT studies ificity of the previous cutoff point were calculated using the present were performed on different scanners at our institution, data. potentially resulting in a small amount of measurement variability that was likely not clinically relevant. Lastly, our AUC CI Cutoff Sensitivity Specificity algorithm was developed based on a limited patient pop- SUV 0.85 0.73–0.96 5.3 91.2 70.0 max ulation at a single institution, so external validation will be SUV 0.83 0.71–0.96 4.5 91.7 65.0 peak needed. TLR 0.84 0.71–0.97 3.0 91.3 68.4 4.1. Clinical Findings. Traditional teaching states that most associated with significant pain [8], but we are unaware of plexiform neurofibromas are asymptomatic, unless trau- anyevidencesupportingthis.Inourpatientpopulation,pain matized or compressed, while MPNSTs are usually and growth were more sensitive and nerve symptoms were 6 Sarcoma NF1 mass Superficial Deep Likely not malignant Possibly malignant NPV 100.0% PPV 61.0% Obtain MRI and PET SUV peak > 4.5 Any necrosis SUV peak > 4.5 Necrosis > 25% SUV peak ≤ 4.5 and any necrosis Sens 87.5% Sens 94.7% Sens 95.2% and no necrosis Sens 56.2% Spec 76.1% Spec 70.8% Spec 75.0% Spec 93.9% Likely not Possibly Likely Very likely Very likely malignant malignant malignant malignant malignant NPV 81.6% PPV 51.4% PPV 72.0% PPV 81.8% PPV 92.3% Biopsy Biopsy Negative Positive Negative Observation Close observation/wide Neoadjuvant chemotherapy/ Rebiopsy/ excision if symptomatic radiation/wide excision wide excision Figure 2: Suggested algorithm for the evaluation and management of PNSTconcerning for malignancy in patients with NF1. It should be notedthatallpredictivevaluesarebasedsolelyonourpatientpopulation(forconfidenceintervals,pleaserefertoTable2)andthealgorithm therefore requires further validation. more specific for diagnosing MPNST. However, either in subcutaneous neurofibromas are often symptomatic but isolation or combination with one another, clinical findings very rarely malignant [11, 29]. &erefore, our recommen- were not as predictive as imaging. dation would be to observe subcutaneous neurofibromas andconsiderexcisioniftheyaregrowingorsymptomatic.In our data, tumor depth had poor specificity for malignancy 4.2. MRI Findings. In the AJCC staging system, sarcomas (21%), consistent with prior studies [30]. greater than 5cm in diameter are considered at higher risk Necrosis, which often results from rapid tumor growth, for local progression and metastasis. &is is generally ac- generally indicates aggressive behavior in sarcomas. &e cepted and well supported in the MPNST literature. Of note, FrenchorFNCLCCsystemutilizeshistologicnecrosis,along the most recent AJCC Cancer Staging Manual removed with differentiation and mitoses, to define tumor grade depth of tumor notation from the guidelines; however, [31, 32]. MRI findings of necrosis have also been associated assessment of tumor depth still remains prominent in the with MPNST [12]. Consistent with this, necrosis visualized literature [26]. &is is generally accepted and well supported on MRI was highly predictive of malignancy in our pop- in the MPNST literature. Kar et al. [27] found that 92% of ulation. Necrosis greater than 25% was the most specific test malignant tumors were >5cm and deep to the fascia, while for malignancy in our study (specificity 95.2%; sensitivity Hwang et al. [28] reported that 57% of malignant tumors 75.0%), while any necrosis was less specific but sensitive were>5cm and 88% were deep to the fascia. We found that (specificity 76.1%; sensitivity 87.5%). In addition, necrosis increasing tumor size was predictive of malignancy as a was significant in several iterations of the regression model. continuous variable in the regression analysis, but the 5cm &us, necrosis on MRI may be an indication for biopsy; AJCC cutoff was neither sensitive nor specific. In addition, furthermore, if histologic necrosis is noted in an otherwise ROC analysis did not identify a clear size above which tu- nondiagnostic tissue specimen, rebiopsy or wide excision of mors were more likely to be malignant, which limits the the tumor should be strongly considered. clinical utility of this variable in diagnosing MPNST. All of our malignant tumors were deep to the fascia, resulting in sensitivity and NPV of approximately 100%. Our patient 4.3. PET Findings. SeveralsemiquantitativemeasuresofPET population did not include any superficial MPNSTs. Con- images have been evaluated and compared for the diagnosis sistent with this, prior literature suggests that cutaneous of MPNST. A review by Treglia et al. found FDG-PET/CTto neurofibromas do not have malignant potential, and be a highly sensitive noninvasive method to identify Sarcoma 7 malignant change in NF1 tumors [24]. SUV has been max Conflicts of Interest widely used, and most studies report thresholds of <2.5 for &e authors declare no conflicts of interest regarding the benign and>3.5 for malignant lesions;however, the rangeof publication of this article. 2.5–3.5 remains indeterminate [24, 33]. Salamon et al. [34] found SUV threshold of>3.5 was sensitive but produced max Acknowledgments a relatively high rate of false positives, while TLR was more specific with a threshold>2.6. SUV has also been peak &e authors would like to thank Dr. Angela Hirbe for studied, with benign tumors ranging from 0.72–3.04 and sharing her expertise in diagnosis and treatment of MPNST, malignant tumors from 2.41–23.38 [23]. Our analysis as well as Carrie Heineman for her assistance with manu- resulted in optimal cutoff values that were comparable to script preparation. thoseinthepriorliteraturefortheseparameters(Table3).In contrast to other studies, we found similar predictive References properties among the quantitative PET measures, with no advantage of TLR over SUV and SUV (Table 2). peak max [1] M. Carli, A. Ferrari, A. Mattke et al., “Pediatric malignant Furthermore, a recent study combined PET/MRI and found peripheral nerve sheath tumor: the Italian and German soft similar results, with the benefits of less radiation and su- tissue sarcoma cooperative group,” Journal of Clinical On- perior imaging than combined PET/CT. &is may be an- cology, vol. 23, no. 33, pp. 8422–8430, 2005. [2] D. G. R. Evans, M. E. Baser, J. McGaughran, S. Sharif, other possible modality for the future [35]. E. Howard, and A. Moran, “Malignant peripheral nerve Qualitative interpretation of PET imaging has also been sheath tumours in neurofibromatosis 1,” Journal of Medical studied in the context of PNST. In a study by Fischer et al., Genetics, vol. 39, no. 5, pp. 311–314, 2002. lesions with increased uptake were identified and rated on a [3] R. E. Ferner, “Neurofibromatosis 1,” European Journal of five-pointvisual scale.&eyconcludedthat PETimagingcan Human Genetics, vol. 15, no. 2, pp. 131–138, 2007. predictgrowthofPNSTbutdidnotexaminetherelationship [4] R. E.Ferner,“Neurofibromatosis 1and neurofibromatosis2:a between growth and malignancy [19]. Chirindel et al. also twenty first centuryperspective,” 7e Lancet Neurology, vol.6, performed a qualitative analysis of PET studies in which no. 4, pp. 340–351, 2007. lesions were visually assessed and dichotomized as either [5] R. E. Ferner and D. H. Gutmann, “International consensus suspected malignant or benign. On early images (performed statement on malignant peripheral nerve sheath tumors in at 1 hour after FDG administration) the sensitivity, speci- neurofibromatosis,” Cancer Research, vol. 62, no. 5, pp. 1573–1577, 2002. ficity, PPV, and NPV were 91%, 84%, 67%, and 96%, re- [6] R. E. Ferner, S. M. Huson, N. &omas et al., “Guidelines for spectively, which are comparable to our findings (94%, 71%, the diagnosis and management of individuals with neurofi- 70%, and 94%, respectively) [18]. &ese results suggest that bromatosis 1,” Journal of Medical Genetics, vol. 44, no. 2, qualitative PET measures may be as accurate as semi- pp. 81–88, 2007. quantitative measures in diagnosing MPNST. [7] J. M. Friedman, “Epidemiology of neurofibromatosis type 1,” American Journal of Medical Genetics, vol. 89, no. 1, pp. 1–6, 5. Conclusion [8] J. H. Tonsgard, “Clinical manifestations and management of neurofibromatosis type 1,” Seminars in Pediatric Neurology, Our results confirmed that larger tumors are more likely to vol. 13, no. 1, pp. 2–7, 2006. be malignant, although we found no clinically relevant size [9] R. Listernick and J. Charrow, “Neurofibromatosis type 1 in threshold, and that MPNSTs superficial to the fascia are childhood,” Journal of Pediatrics, vol.116, no. 6, pp. 845–853, extremely rare. Metabolic measurements were relatively accurate and comparable to one another diagnostically. [10] B. S. Ducatman, B. W. Scheithauer, D. G. Piepgras, Clinical symptoms may also be helpful, although the limi- H.M.Reiman,andD.M.Ilstrup,“Malignantperipheralnerve tations of their predictive properties should be understood sheath tumors: a clinicopathologic study of 120 cases,” and taken into consideration. Finally, necrosis may be a Cancer, vol. 57, no. 10, pp. 2006–2021, 1986. valuable and thus far underutilized predictor of malignancy. [11] T. Tucker, P. Wolkenstein, J. Revuz, J. Zeller, and J. M. Friedman, “Association between benign and malignant peripheral nerve sheath tumors in NF1,” Neurology, vol. 65, Data Availability no. 2, pp. 205–211, 2005. [12] J. Wasa, Y. Nishida, S. Tsukushi et al., “MRI features in the &e datasets generated and analyzed during the current differentiation of malignant peripheral nerve sheath tumors studyarenotpubliclyavailableforprivacyofthesubjectsbut and neurofibromas,” American Journal of Roentgenology, may be available from the corresponding author on rea- vol. 194, no. 6, pp. 1568–1574, 2010. sonable request. [13] S. M. Broski, G. B. Johnson, B. M. Howe et al., “Evaluation of F-FDG PET and MRI in differentiating benign and ma- lignant peripheral nerve sheath tumors,” Skeletal Radiology, Ethical Approval vol. 45, no. 8, pp. 1097–1105, 2016. [14] S. Demehri, A. Belzberg, J. Blakeley, and L. M. Fayad, &is study was approved by Washington University In- “Conventional and functional MR imaging of peripheral stitutional Review Board. &e study was done in accordance nerve sheath tumors: initial experience,” American Journal of with the Declaration of Helsinki. Neuroradiology, vol. 35, no. 8, pp. 1615–1620, 2014. 8 Sarcoma [15] R. E. Friedrich, L. Kluwe, C. Funsterer, ¨ and V. F. Mautner, different clinical features associated with neurofibromatosis “Malignant peripheral nerve sheath tumors (MPNST) in type 1,” Cancer Research and Treatment, vol. 49, no. 3, neurofibromatosis type 1 (NF1): diagnostic findings on pp. 717–726, 2017. [29] F. J. Rodriguez, A. L. Folpe, C. Giannini, and A. Perry, magnetic resonance images and mutation analysis of the NF1 gene,” Anticancer Research, vol. 25, no. 3, pp. 1699–1702, “Pathology of peripheral nerve sheath tumors: diagnostic overview and update on selected diagnostic problems,” Acta Neuropathologica, vol. 123, no. 3, pp. 295–319, 2012. [16] V. F. Mautner, R. E. Friedrich, A. von Deimling et al., [30] D. Furniss, M. C. Swan, D. G. Morritt et al., “A 10-year review “Malignant peripheral nerve sheath tumours in neurofibro- of benign and malignant peripheral nerve sheath tumors in a matosis type 1: MRI supports the diagnosis of malignant single center: clinical and radiographic features can help to plexiform neurofibroma,” Neuroradiology, vol. 45, no. 9, differentiate benign from malignant lesions,” Plastic and pp. 618–625, 2003. Reconstructive Surgery, vol. 121, no. 2, pp. 529–533, 2008. [17] J.Salamon,V.Mautner,G.Adam,andT.Derlin,“Multimodal [31] AJCC Cancer Staging Manual (American Joint Committee on imaging in neurofibromatosis type 1-associated nerve sheath Cancer), Soft Tissue Sarcoma of the Trunk and Extremities, tumors,” RoFo—Fortschritte ¨ auf dem Gebiet der AJCC Cancer Staging Manual, New York, NY, USA, 2017. Rontgenstrahlen ¨ und der Bildgebenden Verfahren, vol. 187, [32] A. Neuville, F. Chibon, and J.-M. Coindre, “Grading of soft no. 12, pp. 1084–1092, 2015. tissue sarcomas: from histological to molecular assessment,” [18] A. Chirindel, M. Chaudhry, J. O. Blakeley, and R. Wahl, “ F- Pathology, vol. 46, no. 2, pp. 113–120, 2014. FDG PET/CT qualitative and quantitative evaluation in [33] R. E. Ferner, J. F. Golding, M. Smith et al., “[ F]2-fluoro-2- neurofibromatosis type 1 patients for detection of malignant deoxy-D-glucose positron emission tomography (FDG PET) transformation: comparison of early to delayed imaging with as a diagnostic tool for neurofibromatosis 1 (NF1) associated and without liver activity normalization,” Journal of Nuclear malignant peripheral nerve sheath tumours (MPNSTs): a Medicine, vol. 56, no. 3, pp. 379–385, 2015. long-term clinical study,” Annals of Oncology, vol. 19, no. 2, [19] M. J. Fisher, S. Basu, E. Dombi et al., “&e role of [ F]- pp. 390–394, 2008. fluorodeoxyglucose positron emission tomography in pre- [34] J. Salamon, S. Veldhoen, I. Apostolova et al., “ F-FDG PET/ dicting plexiform neurofibroma progression,” Journal of CT for detection of malignant peripheral nerve sheath tu- Neuro-Oncology, vol. 87, no. 2, pp. 165–171, 2008. mours in neurofibromatosis type 1: tumour-to-liver ratio is [20] M. A. Bredella, M. Torriani, F. Hornicek et al., “Value of PET superior to an SUVmax cut-off,” European Radiology, vol. 24, in the assessment of patients with neurofibromatosis type 1,” no. 2, pp. 405–412, 2014. American Journal of Roentgenology, vol. 189, no. 4, pp. 928– [35] C. P. Reinert, M. U. Schuhmann, B. Bender et al., “Com- 935, 2007. prehensive anatomical and functional imaging in patients [21] S. Cardona, M. Schwarzbach, U. Hinz et al., “Evaluation of with type I neurofibromatosis using simultaneous FDG-PET/ F18-deoxyglucosepositronemissiontomography(FDG-PET) MRI,” European Journal of Nuclear Medicine and Molecular to assess the nature of neurogenic tumours,” European Imaging, vol. 46, no. 3, pp. 776–787, 2019. Journal of Surgical Oncology, vol. 29, no. 6, pp. 536–541, 2003. [22] A.A.Garsa,A.J.Chang,T.DeWeesetal.,“Prognosticvalueof F-FDG PET metabolic parameters in oropharyngeal squa- mous cell carcinoma,” Journal of Radiation Oncology, vol. 2, no. 1, pp. 27–34, 2013. [23] G. J. R. Cook, E. Lovat, M. Siddique, V. Goh, R. Ferner, and V.S.Warbey,“Characterisationofmalignantperipheralnerve sheath tumours in neurofibromatosis-1 using heterogeneity analysis of F-FDG PET,” European Journal of Nuclear Medicine and Molecular Imaging, vol. 44, no. 11, pp. 1845– 1852, 2017. [24] G. Treglia, S. Taralli, F. Bertagna et al., “Usefulness of whole- body fluorine-18-fluorodeoxyglucose positron emission to- mography in patients with neurofibromatosis type 1: a sys- tematic review,” Radiology Research and Practice, vol. 2012, Article ID 431029, 9 pages, 2012. [25] J. J. Cerci, E. Tabacchi, M. Bogoni et al., “Comparison of CT and PET/CT for biopsy guidance in oncological patients,” European Journal of Nuclear Medicine and Molecular Imaging, vol. 44, no. 8, pp. 1269–1274, 2017. [26] M. B. Amin, F. L. Greene, S. B. Edge et al., “&e eighth edition AJCC cancer staging manual: continuing to build a bridge from a population-based to a more “personalized” approach to cancer staging,” CA: A Cancer Journal for Clinicians, vol. 67, no. 2, pp. 93–99, 2017. [27] M.Kar,S.S.Deo,N.Shuklaetal.,“Malignantperipheralnerve sheath tumors (MPNST)—clinicopathological study and treatment outcome of twenty-four cases,” World Journal of Surgical Oncology, vol. 4, no. 1, p. 55, 2006. [28] I. K. Hwang, S. M. Hahn, H. S. Kim et al., “Outcomes of treatment for malignant peripheral nerve sheath tumors: MEDIATORS of INFLAMMATION The Scientific Gastroenterology Journal of World Journal Research and Practice Diabetes Research Disease Markers Hindawi Hindawi Publishing Corporation Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 http://www www.hindawi.com .hindawi.com V Volume 2018 olume 2013 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 International Journal of Journal of Immunology Research Endocrinology Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Submit your manuscripts at www.hindawi.com BioMed PPAR Research Research International Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Journal of Obesity Evidence-Based Journal of Journal of Stem Cells Complementary and Ophthalmology International Alternative Medicine Oncology Hindawi Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2013 Parkinson’s Disease Computational and Behavioural Mathematical Methods AIDS Oxidative Medicine and in Medicine Neurology Research and Treatment Cellular Longevity Hindawi Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Sarcoma Hindawi Publishing Corporation

How Effective Are Noninvasive Tests for Diagnosing Malignant Peripheral Nerve Sheath Tumors in Patients with Neurofibromatosis Type 1? Diagnosing MPNST in NF1 Patients

Download PDF
9 pages

Loading next page...
 
/lp/hindawi-publishing-corporation/how-effective-are-noninvasive-tests-for-diagnosing-malignant-agI6xtF8OH

References (35)

Publisher
Hindawi Publishing Corporation
Copyright
Copyright © 2019 Maria Schwabe et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ISSN
1357-714X
eISSN
1369-1643
DOI
10.1155/2019/4627521
Publisher site
See Article on Publisher Site

Abstract

Hindawi Sarcoma Volume 2019, Article ID 4627521, 8 pages https://doi.org/10.1155/2019/4627521 Research Article How Effective Are Noninvasive Tests for Diagnosing Malignant Peripheral Nerve Sheath Tumors in Patients with Neurofibromatosis Type 1? Diagnosing MPNST in NF1 Patients Maria Schwabe , Stanislav Spiridonov, Elizabeth L. Yanik, Jack W. Jennings, Travis Hillen, Maria Ponisio, Douglas J. McDonald, Farrokh Dehdashti, and Cara A. Cipriano Washington University School of Medicine in St. Louis, 660 South Euclid Avenue, St. Louis, MO 63110, USA Correspondence should be addressed to Cara A. Cipriano; cipriano@wustl.edu Received 20 February 2019; Accepted 30 May 2019; Published 1 July 2019 Academic Editor: Manish Agarwal Copyright©2019MariaSchwabeetal.&isisanopenaccessarticledistributedundertheCreativeCommonsAttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Background. Distinguishing between benign and malignant peripheral nerve sheath tumors (MPNSTs) in neurofibromatosis 1 (NF1) patients prior to excision can be challenging. How can MPNST be most accurately diagnosed using clinical symptoms, magnetic resonance imaging (MRI) findings (tumor size, depth, and necrosis), positron emission tomography (PET) measures (SUV , SUV , SUV /SUV , and qualitative scale), and combinations of the above? Methods. All NF1 patients peak max max tumor mean liver who underwent PET imaging at our institution (January 1, 2007–December 31, 2016) were included. Medical records were reviewedforclinicalfindings;MRimagesandPETimageswereinterpretedbytwofellowship-trainedmusculoskeletalandnuclear medicine radiologists, respectively. Receiver operating characteristic (ROC) curves were created for each PETmeasurement; the areaunderthecurve(AUC)and thresholdsfordiagnosing malignancywere calculated.Logisticregressiondeterminedsignificant predictors of malignancy. Results. Our population of 41 patients contained 34 benign and 36 malignant tumors. Clinical findings did not reliably predict MPNST. Tumor depth below fascia was highly sensitive; larger tumors were more likely to be malignant but without a useful cutoff for diagnosis. Necrosis on MRI was highly accurate and was the only significant variable in the regression model. PETmeasures were highly accurate, with AUCs comparable and cutoff points consistent with prior studies. A diagnostic algorithm was created using MRI and PET findings. Conclusions. MRI and PET were more effective at diagnosing MPNST than clinical features. We created an algorithm for preoperative evaluation of peripheral nerve sheath tumors in NF1 patients, for which additional validation will be indicated. Differentiating between benign and malignant PNSTs 1. Introduction can be challenging, especially in individuals with multiple Neurofibromatosis type 1 (NF1) is one of the most common neurofibromas. Traditionally, this has been attempted based autosomal-dominant diseases worldwide [1–8]. It has an in- on imaging characteristics and symptoms, which may in- clude pain, increasing size of a mass, and new neurological cidenceof1/2,500to1/3,500individuals[1–5,8,9]andiscaused bymutationsoftheNF1 genelocatedonchromosome17q11.2. deficit [5]. However, there is significant overlap in the ap- Clinically, the disease is characterized by multiple plexiform pearance as well as clinical manifestations of benign and neurofibromas that are usually benign; however, they have the malignant tumors [5, 12]. potential for malignant transformation, with a lifetime risk of Several magnetic resonance imaging (MRI) features can 8–12% [1–5, 8]. Malignant peripheral nerve sheath tumors be useful in distinguishing MPNSTs from neurofibromas. (MPNSTs) are the leading cause of mortality in NF1, reducing &ese include largest dimension of the mass, heterogeneity average life expectancy by 10–15 years [2, 6, 10, 11]. indicating necrosis, peripheral enhancement pattern, 2 Sarcoma perilesional edema-like zone, intratumoral cystic lesion, and on T1 sequences and analyzed as both a continuous and irregular margins [12, 13]. Of these features, tumor size and categorical variable, thelatter using5cmas acutoffbased on necrosis are the best supported predictors for diagnosing the AJCC staging system. Tumor depth was analyzed as a MPNST [12, 14–17] (Figure 1(a)). categorical variable, either superficial or deep to the fascia. Multiple efforts have been made to accurately diagnose Necrosis was defined as nonenhancement on T1 fat- malignant transformation using metabolic imaging with saturated postcontrast images, often with increased T2 positron emission tomography/computed tomography signal intensity, and recorded in quartiles (0%, <25%, 25– (PET/CT)and[ F]fluorodeoxyglucose(FDG)(Figure1(b)). 49%, 50–74%, and ≥75% necrosis). Several authors have studied both semiquantitative and qualitative methods of evaluating lesions with overall good 2.3. FDG-PET/CT Protocol. Patients were imaged according success [8, 18, 19]. Parameters for semiquantitative analysis to our institution’s standard protocol for FDG-PET/CT, include but are not limited to mean standardized uptake which has previously been described [22]. In brief, pa- value(SUV ),maximumSUV(SUV ),maximumSUV mean max tients fasted for at least 4 hours prior to FDG injection. corrected for lean body mass (SUL ), and various ratios max Bloodglucoselevelswererequiredtobe≤200mg/dLpriorto comparing tumor FDG avidity to that of other tissues, such FDG administration. Patients were injected with approxi- as the liver, muscle, and fat. Among the most common of mately 10–15mCi (370–555MBq) of FDG. &e PET/CT these ratios is SUV /SUV , also referred to as max tumor mean liver acquisition was started approximately 60 minutes after the tumor-to-liver ratio (TLR). Semiquantitative methods FDG injection. Patients were scanned from the base of the for diagnosing MPNST, such as mean SUV , have sen- max skull to the upper-thigh with extension to extremities based sitivities of 94%–100% and specificities of 76%–94% [18]. on the location of the lesion of the interest. Noncontrast CT Similarly, qualitative methods, such as visual descriptions of images were obtainedfirst for attenuation correction and for hypermetabolic lesions, have yielded sensitivities of 91%– fusion with the PET images for lesion localization. PET 100% and specificities of 67%–95% [18, 20, 21]. images were acquired at typically 6–8 bed positions, with an In spite of thisresearch, to our knowledge, a noninvasive acquisition time of 2–5 minutes per bed position. gold standard algorithm for diagnosing MPNST has not been established, nor have imaging modalities been evalu- ated in combination with clinical features. &e aim of our 2.4. Semiquantitative Analysis. SUVs were measured using retrospective study was to develop a strategy for dis- the commercial software Hermes (Hermes Medical Solu- tinguishing benign from malignant PNSTusing noninvasive tions, Sweden) by placing a volume of interest (VOI) with a observations and tests, specifically clinical symptoms, MRI diameter of 1.5cm over the most intense region of the lesion features (size, depth, and necrosis), PETmeasures (SUV , on the axial PET/CT images. For SUV , the highest SUV peak max SUV , SUV /SUV , and qualitative scale), and for SUV , the average of the highest SUV within the max max tumor mean liver peak and combinations of the above. VOIs were measured and recorded. When necessary, the diameter of the sphere was adjusted to accommodate for lesion size. Mean liver SUV (average SUV within the VOIs) 2. Materials and Methods was measured using the same 1.5cm diameter sphere placed over the right lobe of the liver. TLR was calculated using 2.1. Patient Population and Data Collection. Following IRB SUV of the tumor over SUV of the liver. approval, all patients with a diagnosis of NF1 who were max mean treated at our institution between 1 January 2007 and 31 December 2016 were identified by searching the medical 2.5. Qualitative Analysis. FDG-PET/CT images were eval- oncology, nuclear medicine, and surgical databases at our uated by one nuclear medicine physician and one nuclear institution. Our study included all patients who underwent medicine fellow who had completed a diagnostic radiology FDG-PET/CT to evaluate for potential malignant trans- residency. Sites of abnormal metabolic activity were scored formation of a PNST with available imaging and confirmed on a 5-point scale based on the following criteria: score of histopathology. Electronic medical records were reviewed for 1 �uptake similar to background, score of 2 �uptake greater demographic information (patient age at time of surgery, than background but less than mediastinal blood pool gender, andtumorlocation)as wellas potential predictors for (MBP), score of 3 �uptake>MBP but less than or equal to malignancy. &ese included preoperative MRI features (tu- liver, score of 4 uptake>liver, and score of 5 �uptake morsize,tumordepthrelativetothefascia,andnecrosis)[22], markedly>than liver (greater than 2-3 times). PETimaging measures (SUV , SUV , and SUV / max peak max tumor SUV ), and clinical findings (pain, enlargement, and mean liver 2.6. Statistical Analysis. Tumor necrosis was analyzed as a nerve symptoms). Histopathology results from biopsy or categorical variable, once with any necrosis considered a surgery obtained through chart review were used as the gold positive finding and once with necrosis >25% considered a standard for diagnosing benign versus malignant PNST. positive finding. For each PET measurement, receiver op- erating characteristic curves (ROC) were created, and the 2.2. MRI Analysis. MR images were assessed by two areaunderthecurve(AUC)wascalculated,alongwithcutoff fellowship-trained musculoskeletal radiologists blinded to points for diagnosing malignancy that optimized sensitivity diagnoses. Tumor size was measured as the largest diameter and specificity. When these were conflicting, sensitivity was Sarcoma 3 5.5 (a) (b) Figure 1: Example of a malignant peripheral nerve sheath tumor in the adductor musculature of the left thigh visualized using (a) magnetic resonance imaging (MRI) and (b) positron emission tomography (PET) studies. &e hypointense areas marked with the asterisk represent necrosis within the tumor. prioritized in order to minimize false negative results in the cases (51.4%). Most patients reported pain and enlargement context of evaluating for malignancy. &ese cutoff points of tumors but not associated nerve symptoms. Additional andtheirsensitivity/specificityinourdatasetwerecompared descriptive statistics including demographic information to cutoff points previously reported in the literature. &e and potential predictors of malignancy are summarized in Table 1. sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated for each tumor characteristic, imaging measure, and clinical finding, 3.1. Clinical Findings. Clinical findings were available in 41 and relevant combinations of these variables. Logistic re- gression was used to evaluate variables for predicting ma- patients for 60 of the 70 tumors. Among patient symptoms, we found that pain and enlargement were more sensitive lignancy in combination. Superficial tumors were not included in the model because no malignancies were ob- (sensitivity 85.2% and 85.2%; specificity 28.1% and 25.0%, respectively), while nerve symptoms were more specific served within this group. Potential predictors (tumor size, necrosis, PET parameters, and clinical findings) were se- (sensitivity 44.8%; specificity 75.0%). Combining nerve symptoms and either pain or growth did not improve di- lected for entry into the model based on the results of the accuracy analysis above. agnostic accuracy compared to evaluation of nerve symp- toms alone. Of note, having any two of three symptoms was more specific and slightly more sensitive than having any 2.7. Diagnostic Algorithm. Our aggregate results were used one particular symptom. Having all three symptoms was to create an algorithm for suggested evaluation of PNST in insensitive and only moderately specific and therefore not NF1 patients. In order to develop a clinically relevant diagnostically useful (Table 2). workup strategy, tumors were first differentiated by tumor features that can be determined by history and physical examination, then by noninvasive imaging studies, and fi- 3.2. MRI Findings. Fifty-nine lesions had PET/CTimages, of nally by biopsy. For each branch point, we calculated the which 37 had MRI images available for review. Tumor depth NPV and PPV to inform clinicians about the likelihood of below the fascia was highly sensitive for malignancy (100%) malignancy given the available information. Lastly, we used with 100% NPV, but not specific (20.7%). Increasing tumor the PPV from the noninvasive workup to offer recom- size was predictive of malignancy as a continuous variable in mendations for management following biopsy, taking pre- regressionanalysis;however,thecutoffof5cm(basedonthe test probability into account. AJCC system) was neither sensitive nor specific. An ROC curve was also constructed for tumor size, but with AUC 0.687 (CI 0.539–0.835), there was no clear threshold for 3. Results diagnosing malignancy. Between the two musculoskeletal radiologists reviewing Our population of 41 patients contained 34 benign and 36 malignanttumors.&e mean patientagewas 30 yearsoldfor MRI studies, interobserver agreement was good (kappa the overall population (range 9–62). Within the 41 patients, 0.608, 95% CI 0.409–0.807) for necrosis by quartiles (none, there was a predominance of tumors in females (41) com- 0–25%, 25–50%, etc.). Agreement was very good for no pared to males (29). &e majority of tumors were deep and necrosis versus any necrosis (kappa 0.937, 95% CI: 0.816– 1.000) and necrosis less than versus greater than 25% (kappa axially located, and the mean diameter was 6.5cm (range 1.5–20.0cm). Mean SUV , SUV , and TLR were 7.9 0.852, 95% CI: 0.655–1.000). For diagnosing MPNST, any max peak necrosis seen on MRI was sensitive (87.5%) and fairly (standard deviation (SD)±5.4), 6.4 (SD±4.3), and 4.0 (SD±2.6), respectively. Necrosis was evident on MRI in 19 specific (76.1%), whereas necrosis of >25% of the tumor SUV 4 Sarcoma Table 1: Descriptive statistics and potential predictors of malig- findings on either MRI (any necrosis) or PET nancy for the 70 peripheral nerve sheath tumors in our study (SUV >4.5) were highly sensitive (95.5%), while posi- peak population of 41 NF1 patients. &ese included patient character- tive findings on both MRI and PET were highly specific istics (age and gender), PET findings (SUV , SUV , and TLR), max peak (93.9%). MRI findings (size, depth relative to the fascia, and necrosis) and Logistic regression analysis supported the value of ne- clinical findings (pain, tumor enlargement, and nerve symptoms), crosis and PETmeasures for diagnosing MPNST. Depth was and histologic diagnosis. found to have 100% sensitivity, with all malignancies oc- Mean 30 curring deep to the fascia in our population; therefore, only Age (years) Median 28 Patient 27 deep tumors were included for the development of Range 9–62 characteristics prediction models. In order to prevent multicollinearity Gender (# of tumors Female 41 between similar measures, SUV >4.5 was selected to each) Male 29 peak represent PET measures, and necrosis >25% of the tumor Yes 46 Pain volume was chosen to represent MRI findings. &us, tumor No 14 size, necrosis >25%, and SUV were selected as the rel- Yes 46 peak Clinical findings Enlargement evant variables; when these were entered for deep tumors, No 14 Yes 23 necrosis was the only significant predictor (p � 0.033, Nerve symptoms No 37 Nagelkerke R �0.622). Mean 6.5 Size (cm) Median 5.4 Range 1.5–20.0 3.4. Combined Diagnostic Algorithm. Clinical findings were MRI findings Depth (relative to Superficial 7 not included in the algorithm as they could not reliably rule fascia) Deep 63 out malignancy in our population (NPV 62.3–69.2%). Yes 19 SUV was included because it had the highest sensitivity Necrosis peak No 18 and NPV of the semiquantitative PETparameters; however, Mean 7.9 the visual scale was almost identical, and the other PET SUV Median 7.4 max measures were not significantly different, so these could be Range 0.7–22.6 substituted for SUV with minimal effect. peak Mean 6.4 Tumors were first differentiated by depth relative to the PET findings SUV Median 5.0 peak fascia, which was notable for 100% NPV in our analysis. Range 0.5–18.7 While tumor depth may be verified on MRI (as in our Mean 4.0 methods), in most cases, it can be easily determined by TLR Median 3.7 Range 0.4–11.9 physical examination prior to advanced imaging. We therefore suggest that patients appearing to have deep tu- Yes 36 Histology Malignant No 34 mors on physical examination be evaluated with MRI and PET and that they undergo biopsy in the presence of con- cerning features on either of these studies. Tumors that are volume was less sensitive (75.0%) but more specific (95.2%), histologically confirmed to be malignant should be managed with PPV 92.3%. with surgery and neoadjuvant/adjuvant chemotherapy/ radiation according to standard protocols. Unfortunately, biopsy itself is imperfect due to sampling error [25]. For 3.3. PET Findings. &e ROC curves for SUV , SUV , max peak tumors that are likely malignant based on imaging but do and TLR were noted to have similar AUCs (Table 3). Op- not contain evidence of MPNST on initial biopsy, rebiopsy timal cutoff points were chosen to maximize sensitivity and or wide excision may be indicated; in contrast, observation specificity based on our patient population. Our cutoff may be acceptable for tumors that appear less concerning points for SUV (4.5) and TLR (3.0) were identical to peak (Figure 2). Of note, this algorithm is based solely on our those in the prior literature, and our cutoff for SUV was max patient population and will therefore require further similar (5.3 in our data compared to 5.0 in the literature, validation. with some more conservative cutoffs slightly lower) [13, 23, 24]. In our dataset, the previously established cutoff of 5.0 resulted in identical sensitivity but lower specificity 4. Discussion (60% compared to 70%) than our optimal cutoff point of 5.3 (Table 3). &e aim of our study was to report the diagnostic value of For qualitative assessment,the interobserver agreement tumor size and depth, MRI features, PET measures, and was verygood, withkappa0.896, 95%CI 0.740–1.000.Level clinical findings to distinguish between benign and malig- 5/5 on the visual scale was considered suggestive of ma- nant PNST. In sum, PET measures and necrosis on MRI lignancy. Using these thresholds for diagnosing MPNST, were the most predictive of malignancy and can be com- PET measures were comparable to one another, with bined to direct workup and treatment in this challenging generallygoodpredictivevalue.Ofthese,SUV >4.5had clinical situation. &is algorithm has been adopted at our peak the highest sensitivity and NPV and was therefore selected institution and will require validation with long-term for testing in combination with MRI necrosis. Positive follow-up from multiple centers. Sarcoma 5 Table2:Sensitivity,specificity,positivepredictivevalues,andnegativepredictivevalues,alongwiththeirassociatedconfidenceintervals,for potential predictors of malignancy. &ese included tumor characteristics (depth relative to the fascia, and size>5cm in diameter), MRI findings (any necrosis, or necrosis>25% by volume), PET findings (SUV >5, SUV peak>4.5, and TLR>3), and clinical findings (pain, max enlarging, and nerve symptoms). Combinations of imaging findings (SUV >or<4.5 on PET and any or >25% necrosis on MRI) and peak clinical findings (1 of 3, 2 of 3, or 3 of 3 symptoms) are also included. Sensitivity Specificity PPV NPV Accuracy 100.0 20.7 61.0 100.0 Deep 64.6 (88.0–100.0) (8.7–40.3) (47.4–73.2) (51.7–100.0) Tumor characteristics 68.0 53.5 56.7 65.2 Size>5cm 60.4 (46.4–84.2) (34.2–72.0) (37.7–74.0) (42.8–82.8) 87.5 76.1 51.4 73.7 Necrosis (any) 81.1 (60.4–97.8) (52.4–90.9) (34.7–67.8) (32.2–65.3) MRI findings 75.0 95.2 92.3 83.3 Necrosis (>25%) 86.5 (47.4–91.7) (74.1–99.8) (62.1–100) (61.8–94.5) 89.3 73.3 75.8 88.0 SUV >5 78.0 max (70.6–97.2) (50.8–87.0) (57.4–88.2) (67.7–96.8) 94.7 70.8 72.0 94.4 SUV >4.5 78.0 peak (71.9–99.7) (48.8–86.6) (50.4–87.1) (70.6–99.7) PET findings 91.7 73.9 78.6 89.5 TLR>3 82.1 (71.5–98.5) (51.3–88.90 (58.5–91.0) (65.5–98.2) 94.1 70.8 69.6 94.4 Visual score 80.5 (69.2–99.7) (48.7–86.6) (47.0–86.0) (70.6–99.7) SUV >4.5 (PET) or any necrosis 95.5 66.7 67.7 95.2 peak 78.8 (MRI) (75.1–100) (47.1–82.1) (48.5–82.7) (74.1–100) SUV >4.5 (PET) and any necrosis 56.2 93.9 81.8 81.6 peak 81.6 Combined PET/MRI (MRI) (30.6–79.2) (78.4–98.9) (47.8–96.8) (65.1–91.7) findings SUVpeak>4.5(PET)ornecrosis>25% 72.2 50.0 60.1 64 (42.8–81.2) 61.1 (MRI) (46.5–90.3) (26.0–74.0) (36.7–78.5) SUVpeak>4.5 (PET) and necrosis 61.5 100 100 78.3 83.9 >25% (MRI) (31.6–86.14) (81.5–100) (59.8–100) (64.4–87.6) 85.2 28.1 50 69.2 Pain 54.2 (65.4–95.1) (14.4–47.0) (35.1–64.9) (38.9–89.6) 85.2 25.0 48.9 66.7 Clinical findings Growth 52.5 (65.4–95.1) (12.1–43.8) (34.3–63.7) (35.4–88.7) 44.8 75.0 59.1 62.3 Nerve symptoms 61.0 (26.9–64.0) (57.5–87.3) (36.7–78.5) (46.7–76.6) 92.9 9.68 48.1 60.0 1 of 3 findings 49.1 (75.0–98.8) (2.53–26.9) (34.5–62.0) (17.0–92.7) Combined clinical 89.3 43.8 58.1 82.3 2 of 3 findings 65.0 findings (70.6–97.2) (26.8–62.1) (42.2–72.6) (55.8–95.3) 32.1 78.1 56.2 43.2 3 of 3 findings 56.7 (16.6–5.24) (60.0–90.1) (30.6–79.2) (28.7–58.9) Our study had several limitations. It is a retrospective Table 3: Areas under the curve (AUC) and associated confidence analysiswitharelativelysmallsamplesizeduetotherarityof intervals (CI) from receiver operating characteristic curves for NF1 and MPNST. Most importantly, PET/CT, MRI, and SUV , SUV , and tumor-to-liver ratio (TLR). Using these max peak clinical findings were not available for all patients; however, present data, cutoff points were selected to optimize sensitivity and these missing data did not correlate with year, age, or specificity. Our cutoff point for SUV differed slightly from that max established in the previous literature, so the sensitivity and spec- malignancy characteristics. In addition, PET/CT studies ificity of the previous cutoff point were calculated using the present were performed on different scanners at our institution, data. potentially resulting in a small amount of measurement variability that was likely not clinically relevant. Lastly, our AUC CI Cutoff Sensitivity Specificity algorithm was developed based on a limited patient pop- SUV 0.85 0.73–0.96 5.3 91.2 70.0 max ulation at a single institution, so external validation will be SUV 0.83 0.71–0.96 4.5 91.7 65.0 peak needed. TLR 0.84 0.71–0.97 3.0 91.3 68.4 4.1. Clinical Findings. Traditional teaching states that most associated with significant pain [8], but we are unaware of plexiform neurofibromas are asymptomatic, unless trau- anyevidencesupportingthis.Inourpatientpopulation,pain matized or compressed, while MPNSTs are usually and growth were more sensitive and nerve symptoms were 6 Sarcoma NF1 mass Superficial Deep Likely not malignant Possibly malignant NPV 100.0% PPV 61.0% Obtain MRI and PET SUV peak > 4.5 Any necrosis SUV peak > 4.5 Necrosis > 25% SUV peak ≤ 4.5 and any necrosis Sens 87.5% Sens 94.7% Sens 95.2% and no necrosis Sens 56.2% Spec 76.1% Spec 70.8% Spec 75.0% Spec 93.9% Likely not Possibly Likely Very likely Very likely malignant malignant malignant malignant malignant NPV 81.6% PPV 51.4% PPV 72.0% PPV 81.8% PPV 92.3% Biopsy Biopsy Negative Positive Negative Observation Close observation/wide Neoadjuvant chemotherapy/ Rebiopsy/ excision if symptomatic radiation/wide excision wide excision Figure 2: Suggested algorithm for the evaluation and management of PNSTconcerning for malignancy in patients with NF1. It should be notedthatallpredictivevaluesarebasedsolelyonourpatientpopulation(forconfidenceintervals,pleaserefertoTable2)andthealgorithm therefore requires further validation. more specific for diagnosing MPNST. However, either in subcutaneous neurofibromas are often symptomatic but isolation or combination with one another, clinical findings very rarely malignant [11, 29]. &erefore, our recommen- were not as predictive as imaging. dation would be to observe subcutaneous neurofibromas andconsiderexcisioniftheyaregrowingorsymptomatic.In our data, tumor depth had poor specificity for malignancy 4.2. MRI Findings. In the AJCC staging system, sarcomas (21%), consistent with prior studies [30]. greater than 5cm in diameter are considered at higher risk Necrosis, which often results from rapid tumor growth, for local progression and metastasis. &is is generally ac- generally indicates aggressive behavior in sarcomas. &e cepted and well supported in the MPNST literature. Of note, FrenchorFNCLCCsystemutilizeshistologicnecrosis,along the most recent AJCC Cancer Staging Manual removed with differentiation and mitoses, to define tumor grade depth of tumor notation from the guidelines; however, [31, 32]. MRI findings of necrosis have also been associated assessment of tumor depth still remains prominent in the with MPNST [12]. Consistent with this, necrosis visualized literature [26]. &is is generally accepted and well supported on MRI was highly predictive of malignancy in our pop- in the MPNST literature. Kar et al. [27] found that 92% of ulation. Necrosis greater than 25% was the most specific test malignant tumors were >5cm and deep to the fascia, while for malignancy in our study (specificity 95.2%; sensitivity Hwang et al. [28] reported that 57% of malignant tumors 75.0%), while any necrosis was less specific but sensitive were>5cm and 88% were deep to the fascia. We found that (specificity 76.1%; sensitivity 87.5%). In addition, necrosis increasing tumor size was predictive of malignancy as a was significant in several iterations of the regression model. continuous variable in the regression analysis, but the 5cm &us, necrosis on MRI may be an indication for biopsy; AJCC cutoff was neither sensitive nor specific. In addition, furthermore, if histologic necrosis is noted in an otherwise ROC analysis did not identify a clear size above which tu- nondiagnostic tissue specimen, rebiopsy or wide excision of mors were more likely to be malignant, which limits the the tumor should be strongly considered. clinical utility of this variable in diagnosing MPNST. All of our malignant tumors were deep to the fascia, resulting in sensitivity and NPV of approximately 100%. Our patient 4.3. PET Findings. SeveralsemiquantitativemeasuresofPET population did not include any superficial MPNSTs. Con- images have been evaluated and compared for the diagnosis sistent with this, prior literature suggests that cutaneous of MPNST. A review by Treglia et al. found FDG-PET/CTto neurofibromas do not have malignant potential, and be a highly sensitive noninvasive method to identify Sarcoma 7 malignant change in NF1 tumors [24]. SUV has been max Conflicts of Interest widely used, and most studies report thresholds of <2.5 for &e authors declare no conflicts of interest regarding the benign and>3.5 for malignant lesions;however, the rangeof publication of this article. 2.5–3.5 remains indeterminate [24, 33]. Salamon et al. [34] found SUV threshold of>3.5 was sensitive but produced max Acknowledgments a relatively high rate of false positives, while TLR was more specific with a threshold>2.6. SUV has also been peak &e authors would like to thank Dr. Angela Hirbe for studied, with benign tumors ranging from 0.72–3.04 and sharing her expertise in diagnosis and treatment of MPNST, malignant tumors from 2.41–23.38 [23]. Our analysis as well as Carrie Heineman for her assistance with manu- resulted in optimal cutoff values that were comparable to script preparation. thoseinthepriorliteraturefortheseparameters(Table3).In contrast to other studies, we found similar predictive References properties among the quantitative PET measures, with no advantage of TLR over SUV and SUV (Table 2). peak max [1] M. Carli, A. Ferrari, A. Mattke et al., “Pediatric malignant Furthermore, a recent study combined PET/MRI and found peripheral nerve sheath tumor: the Italian and German soft similar results, with the benefits of less radiation and su- tissue sarcoma cooperative group,” Journal of Clinical On- perior imaging than combined PET/CT. &is may be an- cology, vol. 23, no. 33, pp. 8422–8430, 2005. [2] D. G. R. Evans, M. E. Baser, J. McGaughran, S. Sharif, other possible modality for the future [35]. E. Howard, and A. Moran, “Malignant peripheral nerve Qualitative interpretation of PET imaging has also been sheath tumours in neurofibromatosis 1,” Journal of Medical studied in the context of PNST. In a study by Fischer et al., Genetics, vol. 39, no. 5, pp. 311–314, 2002. lesions with increased uptake were identified and rated on a [3] R. E. Ferner, “Neurofibromatosis 1,” European Journal of five-pointvisual scale.&eyconcludedthat PETimagingcan Human Genetics, vol. 15, no. 2, pp. 131–138, 2007. predictgrowthofPNSTbutdidnotexaminetherelationship [4] R. E.Ferner,“Neurofibromatosis 1and neurofibromatosis2:a between growth and malignancy [19]. Chirindel et al. also twenty first centuryperspective,” 7e Lancet Neurology, vol.6, performed a qualitative analysis of PET studies in which no. 4, pp. 340–351, 2007. lesions were visually assessed and dichotomized as either [5] R. E. Ferner and D. H. Gutmann, “International consensus suspected malignant or benign. On early images (performed statement on malignant peripheral nerve sheath tumors in at 1 hour after FDG administration) the sensitivity, speci- neurofibromatosis,” Cancer Research, vol. 62, no. 5, pp. 1573–1577, 2002. ficity, PPV, and NPV were 91%, 84%, 67%, and 96%, re- [6] R. E. Ferner, S. M. Huson, N. &omas et al., “Guidelines for spectively, which are comparable to our findings (94%, 71%, the diagnosis and management of individuals with neurofi- 70%, and 94%, respectively) [18]. &ese results suggest that bromatosis 1,” Journal of Medical Genetics, vol. 44, no. 2, qualitative PET measures may be as accurate as semi- pp. 81–88, 2007. quantitative measures in diagnosing MPNST. [7] J. M. Friedman, “Epidemiology of neurofibromatosis type 1,” American Journal of Medical Genetics, vol. 89, no. 1, pp. 1–6, 5. Conclusion [8] J. H. Tonsgard, “Clinical manifestations and management of neurofibromatosis type 1,” Seminars in Pediatric Neurology, Our results confirmed that larger tumors are more likely to vol. 13, no. 1, pp. 2–7, 2006. be malignant, although we found no clinically relevant size [9] R. Listernick and J. Charrow, “Neurofibromatosis type 1 in threshold, and that MPNSTs superficial to the fascia are childhood,” Journal of Pediatrics, vol.116, no. 6, pp. 845–853, extremely rare. Metabolic measurements were relatively accurate and comparable to one another diagnostically. [10] B. S. Ducatman, B. W. Scheithauer, D. G. Piepgras, Clinical symptoms may also be helpful, although the limi- H.M.Reiman,andD.M.Ilstrup,“Malignantperipheralnerve tations of their predictive properties should be understood sheath tumors: a clinicopathologic study of 120 cases,” and taken into consideration. Finally, necrosis may be a Cancer, vol. 57, no. 10, pp. 2006–2021, 1986. valuable and thus far underutilized predictor of malignancy. [11] T. Tucker, P. Wolkenstein, J. Revuz, J. Zeller, and J. M. Friedman, “Association between benign and malignant peripheral nerve sheath tumors in NF1,” Neurology, vol. 65, Data Availability no. 2, pp. 205–211, 2005. [12] J. Wasa, Y. Nishida, S. Tsukushi et al., “MRI features in the &e datasets generated and analyzed during the current differentiation of malignant peripheral nerve sheath tumors studyarenotpubliclyavailableforprivacyofthesubjectsbut and neurofibromas,” American Journal of Roentgenology, may be available from the corresponding author on rea- vol. 194, no. 6, pp. 1568–1574, 2010. sonable request. [13] S. M. Broski, G. B. Johnson, B. M. Howe et al., “Evaluation of F-FDG PET and MRI in differentiating benign and ma- lignant peripheral nerve sheath tumors,” Skeletal Radiology, Ethical Approval vol. 45, no. 8, pp. 1097–1105, 2016. [14] S. Demehri, A. Belzberg, J. Blakeley, and L. M. Fayad, &is study was approved by Washington University In- “Conventional and functional MR imaging of peripheral stitutional Review Board. &e study was done in accordance nerve sheath tumors: initial experience,” American Journal of with the Declaration of Helsinki. Neuroradiology, vol. 35, no. 8, pp. 1615–1620, 2014. 8 Sarcoma [15] R. E. Friedrich, L. Kluwe, C. Funsterer, ¨ and V. F. Mautner, different clinical features associated with neurofibromatosis “Malignant peripheral nerve sheath tumors (MPNST) in type 1,” Cancer Research and Treatment, vol. 49, no. 3, neurofibromatosis type 1 (NF1): diagnostic findings on pp. 717–726, 2017. [29] F. J. Rodriguez, A. L. Folpe, C. Giannini, and A. Perry, magnetic resonance images and mutation analysis of the NF1 gene,” Anticancer Research, vol. 25, no. 3, pp. 1699–1702, “Pathology of peripheral nerve sheath tumors: diagnostic overview and update on selected diagnostic problems,” Acta Neuropathologica, vol. 123, no. 3, pp. 295–319, 2012. [16] V. F. Mautner, R. E. Friedrich, A. von Deimling et al., [30] D. Furniss, M. C. Swan, D. G. Morritt et al., “A 10-year review “Malignant peripheral nerve sheath tumours in neurofibro- of benign and malignant peripheral nerve sheath tumors in a matosis type 1: MRI supports the diagnosis of malignant single center: clinical and radiographic features can help to plexiform neurofibroma,” Neuroradiology, vol. 45, no. 9, differentiate benign from malignant lesions,” Plastic and pp. 618–625, 2003. Reconstructive Surgery, vol. 121, no. 2, pp. 529–533, 2008. [17] J.Salamon,V.Mautner,G.Adam,andT.Derlin,“Multimodal [31] AJCC Cancer Staging Manual (American Joint Committee on imaging in neurofibromatosis type 1-associated nerve sheath Cancer), Soft Tissue Sarcoma of the Trunk and Extremities, tumors,” RoFo—Fortschritte ¨ auf dem Gebiet der AJCC Cancer Staging Manual, New York, NY, USA, 2017. Rontgenstrahlen ¨ und der Bildgebenden Verfahren, vol. 187, [32] A. Neuville, F. Chibon, and J.-M. Coindre, “Grading of soft no. 12, pp. 1084–1092, 2015. tissue sarcomas: from histological to molecular assessment,” [18] A. Chirindel, M. Chaudhry, J. O. Blakeley, and R. Wahl, “ F- Pathology, vol. 46, no. 2, pp. 113–120, 2014. FDG PET/CT qualitative and quantitative evaluation in [33] R. E. Ferner, J. F. Golding, M. Smith et al., “[ F]2-fluoro-2- neurofibromatosis type 1 patients for detection of malignant deoxy-D-glucose positron emission tomography (FDG PET) transformation: comparison of early to delayed imaging with as a diagnostic tool for neurofibromatosis 1 (NF1) associated and without liver activity normalization,” Journal of Nuclear malignant peripheral nerve sheath tumours (MPNSTs): a Medicine, vol. 56, no. 3, pp. 379–385, 2015. long-term clinical study,” Annals of Oncology, vol. 19, no. 2, [19] M. J. Fisher, S. Basu, E. Dombi et al., “&e role of [ F]- pp. 390–394, 2008. fluorodeoxyglucose positron emission tomography in pre- [34] J. Salamon, S. Veldhoen, I. Apostolova et al., “ F-FDG PET/ dicting plexiform neurofibroma progression,” Journal of CT for detection of malignant peripheral nerve sheath tu- Neuro-Oncology, vol. 87, no. 2, pp. 165–171, 2008. mours in neurofibromatosis type 1: tumour-to-liver ratio is [20] M. A. Bredella, M. Torriani, F. Hornicek et al., “Value of PET superior to an SUVmax cut-off,” European Radiology, vol. 24, in the assessment of patients with neurofibromatosis type 1,” no. 2, pp. 405–412, 2014. American Journal of Roentgenology, vol. 189, no. 4, pp. 928– [35] C. P. Reinert, M. U. Schuhmann, B. Bender et al., “Com- 935, 2007. prehensive anatomical and functional imaging in patients [21] S. Cardona, M. Schwarzbach, U. Hinz et al., “Evaluation of with type I neurofibromatosis using simultaneous FDG-PET/ F18-deoxyglucosepositronemissiontomography(FDG-PET) MRI,” European Journal of Nuclear Medicine and Molecular to assess the nature of neurogenic tumours,” European Imaging, vol. 46, no. 3, pp. 776–787, 2019. Journal of Surgical Oncology, vol. 29, no. 6, pp. 536–541, 2003. [22] A.A.Garsa,A.J.Chang,T.DeWeesetal.,“Prognosticvalueof F-FDG PET metabolic parameters in oropharyngeal squa- mous cell carcinoma,” Journal of Radiation Oncology, vol. 2, no. 1, pp. 27–34, 2013. [23] G. J. R. Cook, E. Lovat, M. Siddique, V. Goh, R. Ferner, and V.S.Warbey,“Characterisationofmalignantperipheralnerve sheath tumours in neurofibromatosis-1 using heterogeneity analysis of F-FDG PET,” European Journal of Nuclear Medicine and Molecular Imaging, vol. 44, no. 11, pp. 1845– 1852, 2017. [24] G. Treglia, S. Taralli, F. Bertagna et al., “Usefulness of whole- body fluorine-18-fluorodeoxyglucose positron emission to- mography in patients with neurofibromatosis type 1: a sys- tematic review,” Radiology Research and Practice, vol. 2012, Article ID 431029, 9 pages, 2012. [25] J. J. Cerci, E. Tabacchi, M. Bogoni et al., “Comparison of CT and PET/CT for biopsy guidance in oncological patients,” European Journal of Nuclear Medicine and Molecular Imaging, vol. 44, no. 8, pp. 1269–1274, 2017. [26] M. B. Amin, F. L. Greene, S. B. Edge et al., “&e eighth edition AJCC cancer staging manual: continuing to build a bridge from a population-based to a more “personalized” approach to cancer staging,” CA: A Cancer Journal for Clinicians, vol. 67, no. 2, pp. 93–99, 2017. [27] M.Kar,S.S.Deo,N.Shuklaetal.,“Malignantperipheralnerve sheath tumors (MPNST)—clinicopathological study and treatment outcome of twenty-four cases,” World Journal of Surgical Oncology, vol. 4, no. 1, p. 55, 2006. [28] I. K. Hwang, S. M. Hahn, H. S. Kim et al., “Outcomes of treatment for malignant peripheral nerve sheath tumors: MEDIATORS of INFLAMMATION The Scientific Gastroenterology Journal of World Journal Research and Practice Diabetes Research Disease Markers Hindawi Hindawi Publishing Corporation Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 http://www www.hindawi.com .hindawi.com V Volume 2018 olume 2013 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 International Journal of Journal of Immunology Research Endocrinology Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Submit your manuscripts at www.hindawi.com BioMed PPAR Research Research International Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 Journal of Obesity Evidence-Based Journal of Journal of Stem Cells Complementary and Ophthalmology International Alternative Medicine Oncology Hindawi Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2013 Parkinson’s Disease Computational and Behavioural Mathematical Methods AIDS Oxidative Medicine and in Medicine Neurology Research and Treatment Cellular Longevity Hindawi Hindawi Hindawi Hindawi Hindawi www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018

Journal

Sarcoma – Hindawi Publishing Corporation

Published: Jul 1, 2019

There are no references for this article.