Professional Documents
Culture Documents
https://doi.org/10.1007/s11060-019-03205-z
CLINICAL STUDY
Abstract
Introduction Measurement of tumor growth rates over time for patients with meningiomas has important prognostic and
therapeutic implications. Our objective was to compare two methods of measuring meningioma volume: (1) the simplified
ellipsoid (ABC/2) method; and (2) perimetric volume measurements using imaging software modules.
Methods Patients with conservatively managed meningiomas for at least 1.5 years were retrospectively identified from the
VCU Brain and Spine Tumor Registry over a 10-year period (2005–2015). Tumor volumes were independently measured
using the simplified ellipsoid and computerized perimetric methods. Intra class correlations (CC) and Bland–Altman analyses
were performed.
Results A total of 26 patients representing 29 tumors were identified. Across 146 images, there were 24 (16%) images that
were non-measurable using standard application commands with the computerized perimetric method. The mean volume
obtained using the ABC/2 and computerized perimetric methods were 3.2 ± 3.4 cm3 and 3.4 ± 3.5 cm3, respectively. The
mean volume difference was 0.2 cm3 (SE = 0.12; p = 0.10) across measurement methods. The concordance correlation coef-
ficient (CCC) between methods was 0.95 (95% CI 0.91, 0.98).
Conclusions There is excellent correlation between the simplified ellipsoid and computerized perimetric methods of volu-
metric analysis for conservatively managed meningiomas. The simplified ellipsoid method remains an excellent method for
meningioma volume assessment and had an advantage over the perimetric method which failed to allow measurement of
roughly one in six tumors on imaging.
Keywords Comparative studies · Meningioma · ABC/2 · Tumor volume · Tumor burden · Magnetic resonance imaging
13
Vol.:(0123456789)
Journal of Neuro-Oncology
modalities. Chang et al. [9] used volumetric analysis to iden- perimetric volume for each corresponding image using the
tify undetected growth in 17 of 29 meningiomas previously VitreaWorkstation™ (Vital Images Inc, Plymouth, MN,
measured using the simplified ellipsoid formula. They con- USA). This software has not been previously validated in
cluded volumetric analysis was more accurate and superior the literature for measurement of meningiomas. Tri-plane
for detecting tumor progression as compared to the simpli- (axial, sagittal, coronal) T1-weighted post-contrast images
fied ellipsoid method, which underestimated growth rates. were analyzed by using a computer-assisted, hand-drawn
Similarly, Zeidman et al. [10] observed higher growth rate technique to outline regions of interest for each tumor. The
by volume using computerized methods than the simplified software segmentation process was thus semi-automatic.
ellipsoid formula and concluded computerized measure- The software then applied computer-detected perimeters
ments were more informative. Xue et al. however, compared within the defined region of interest to estimate the tumor
manual with computerized methods in intracerebral hemor- volume. The software’s computer-detected perimeters could
rhages and found that manual estimation tended to overesti- be further defined through user interaction and user edit-
mate volume in comparison to computerized measurement ing to adjust for non-tumor anatomical findings within the
in 45% of cases [11]. region of interest. The time required for the software to
The objective of the current study was to further examine measure each meningioma was approximately 10 min per
whether volumetric analysis of meningiomas from perimet- imaging study. Figure 2 provides a representative image of
ric measurements are superior to measurements performed the tumor segmentation process using VitreaWorkstation™.
using the simplified ellipsoid volume calculated from
orthogonal diameters. Statistical analysis
13
Journal of Neuro-Oncology
Fig. 1 T1-weighted post gadolinium contrast images demonstrate perpendicular dimension to “A” in the same axial cut. “C” is longest
how to measure the “A”, “B”, and “C” dimensions. “A” is the longest dimension orthogonal to “A” in the sagittal plane
dimension measured in the axial plane, with “B” being the longest
unique meningiomas met inclusion criteria for this study. Correlation between measurement modalities
The mean age at initial MRI imaging was 61.3 ranging from
36 to 92 years with a median of 62 years. There were 25 A total of 122 of the 146 (84%) scans yielded volume meas-
(96%) females and one (4%) male. Twenty-four patients urements using both the perimeter and ellipsoid modalities
(88%) had a single meningioma and three (12%) patients with subjects contributing between 1 and 8 dyads per tumor.
had two meningiomas. The mean duration of follow-up was The VitreaWorkstation™ was unable to measure tumor vol-
5.7 ± 2.6 years (range 1.8–12.0). The study cohort yielded umes for 24 (16%) of the images available using standard
a total of 146 MRI images with 28 baseline images and application commands. The mean volume obtained using the
118 follow-up images. The mean number of image sets per ABC/2 method for calculating the simplified ellipsoid vol-
patient was 5.6 ± 1.8 (range 3–10). The most common tumor ume was 3.2 ± 3.4 cm3, while the mean volume from com-
location was the convexity (41%) followed by falcine (14%) puterized perimetric methods was 3.4 ± 3.5 cm3. Adjusting
and sphenoid wing (14%). This demographic information is for repeated observations within a patient, the mean vol-
summarized in Table 1. ume difference between methods was 0.2 cm3 (SE = 0.12;
13
Journal of Neuro-Oncology
Fig. 2 T1-weighted post gadolinium contrast images demonstrating tumor measurement using the VitreaWorkstation™
p = 0.10). Although the CCC was 0.95 (95% CI 0.91, 0.98) [21], and meningiomas [9, 10, 22, 23]. Meanwhile, the Vit-
between the two methods, the ABC/2 method resulted in a reaWorkstation™ has largely been used in CT applications
higher estimate of the volume in 76% (95% CI 0.62–0.95) [24–38] with limited use in MRI for tumor response in high
of the images (Fig. 3). The difference in the volume meas- grade gliomas [39] and meningiomas [9]. The VitreaWork-
urements tended to be clustered together, so that if one station™ perimetric method was successful in calculating
method was found to produce a larger value than the other, volumes for 84% of images but was unable to provide volu-
future observations would have the same trend (ICC = 0.58, metric data for 16% of images.
p = 0.002). The Bland Altman plot comparing the ABC/2 to For the images that could not be measured using the
the perimetric method is presented in Fig. 4. VitreaWorkstation™, attempts were made to redefine the
perimeters and recalibrate the windowing, but no automated
volume was produced. It remains unclear why the software
Discussion was unable to calculate these tumor volumes. There may
have been insufficient contrast between the tumor and sur-
In this study, we examined whether perimetric measure- rounding parenchyma for the software to appropriately dis-
ments of meningiomas using the VitreaWorkstation™ are tinguish the tumor from adjacent tissue within the region
as accurate as measurements using the simplified ellipsoid of interest. When reviewing the literature, only one article
method. Based on our results, there is a clear correlation addressed the potential drawback of semi-automatic soft-
between the two methods of volumetric analysis. ware tools as compared to manually derived measurements.
The measurements with the VitreaWorkstation™ rely De Jonge et al. [30] compared semi-automatic and manually
upon user-defined perimeters of regions of interest to derive adjusted left ventricular functional measurements derived
automated calculations of volume, while the simplified ellip- from ECG-gated dual-source CT for three separate CT
soid volume is calculated directly from three orthogonal analysis software packages. They acknowledged good cor-
perpendicular axes (A, B, C). Traditionally, the ellipsoid relation between methods but recognized largely different
formula has been used by clinicians for measuring volume outliers with automatic measurements which were suspected
of intracerebral hemorrhage [16–19], but has been expanded to be due to the inability to detect the heart valve when it
to the measurement of gliomas [20], acoustic neuromas is open. Further research will be necessary to clarify why
13
Journal of Neuro-Oncology
Table 1 Characteristics of 26 patients with conservatively managed method would be expected to be less accurate for irregu-
meningiomas larly shaped tumors, we note that the volume obtained by
Variable Study cohort (n = 26) the measurement of an irregularly shaped tumor would
be expected to be representative of the tumor volume at a
Female 25 (96.2%)
particular time, and should be comparable to subsequent
Age at initial imaging (years) 61.3 ± 13.7
measurements of the same tumor over time, allowing for an
Years of follow-up 5.7 ± 2.6
assessment of stability or growth over time by comparing
Number of images 146
the volumes represented by the ABC/2 method.
Mean number of images 6.2 ± 1.9
Locations in our tumor sample primarily consisted of
Number of tumors
convexity (41.4%), falcine (13.8%) and sphenoid wing
1 26 (88%)
(13.8%). This is comparable to prior estimates of 60–70%
2 3 (12%)
arising along the falx, sphenoid bone, or convexity [40].
Tumor location
The sample may over-represent convexity meningiomas,
Convexity 12 (41.4%)
and these are likely slightly overrepresented due to the con-
Falcine 4 (13.8%)
servatively managed nature of this series.
Sphenoid wing 4 (13.8%)
The measurements reported herein were carried out by
Planum sphenoide 2 (6.9%)
one individual, and subsequent studies of tumor growth
Tentorial 2 (6.9%)
(to be reported separately) confirm reproducibility of these
Anterior fossa 1 (3.5%)
measurements. Future studies of interuser reliability of the
Cavernous sinus 1 (3.5%)
measurement methods would also be valuable. A major
Intraventricular 1 (3.5%)
strength of the current study is the potential generalizability
Middle fossa 1 (3.5%)
to the broader population. The data is drawn from a retro-
Posterior fossa 1 (3.5%)
spective sample of all patients with brain tumors presenting
to a large academic institution over a 10-year time period
there are circumstances where the software cannot accu- (2005–2015). Each patient had on average at least 6 sepa-
rately identify the anatomy of interest. rate imaging studies over an average of 5.6 years of follow-
When comparing the methods of meningioma measure- up. This allowed for analysis across a substantial number
ment, there was a clear correlation (CCC: 0.95; 95% CI 0.91, of images for both modalities, albeit from a smaller num-
0.98) between measurements using the VitreaWorkstation™ ber of actual individual tumors. While this does introduce
and simplified ellipsoid methods. This helps demonstrate bias related to repeated measurements of tumors, it was felt
that both methods may be effective for the clinical assess- that the increased sample size available by comparing all
ment of meningioma growth and volume changes over time, independently performed measurements from both methods
as monitoring meningiomas for growth is an important was more valuable than the potential bias noted. The non-
aspect of non-operative management of these tumors. The operative nature of this sample provides no way of histo-
results from the current study cannot directly distinguish logically confirming the diagnosis as a meningioma, so it
which method is more accurate for measuring actual tumor strictly relies upon tumor characteristics from imaging alone
volume. The ease, reliability and reproducibility of the vol- for diagnosis. Despite this limitation, this study supports the
umes produced by the ABC/2 method make this an attrac- widely-practiced extrapolation of volume from diameters as
tive method for general clinical use. Indeed, the authors are a reliable and accurate method for the measurement of men-
preparing to submit results of monitoring the same patient ingiomas. Additionally, ABC/2 is easily performed in clini-
cohort for growth of their tumors over time using the ABC/2 cal settings, even without access to electronic methods. Vit-
method and related measurements. reaWorkstation™ is not routinely used in the measurement
None of the tumors exhibited bony invasion or meningi- of meningiomas as it requires a specialized workstation.
omatosis, and all were homogeneously enhancing on imag- Other limitations of the software include time necessary to
ing. Any evidence of calcification or necrosis would have measure tumor volumes, the level of resources required to
shown non-enhancement and be contained within the tumor. have the software readily available, and the personnel skill
This study demonstrates the equivalence of the simplified and training required to use the software. We note also that
ellipsoid method and VitreaWorkstation™ in measuring a there are other software products available for volumetric
variety of meningiomas, but was not powered adequately to analysis of MRI scan data, which could have other features,
determine whether these methods are equivalent in meas- benefits or potential drawbacks relative to the system used
uring irregularly shaped meningiomas. While the ABC/2 in the current study.
13
Journal of Neuro-Oncology
Fig. 3 Comparisons of ABC/2
and perimetric methods of
tumor volume measurement.
A line of accuracy is superim-
posed on the plot. The symbols
indicate unique dyads
13
Journal of Neuro-Oncology
16% of images. While other investigators suggest perimet- 10. Zeidman LA, Ankenbrandt WJ, Du H, Paleologos N, Vick NA
ric methods are preferable, these results demonstrate that (2008) Growth rate of non-operated meningiomas. J Neurol
255(6):891–895. https://doi.org/10.1007/s00415-008-0801-2
the simplified ellipsoid method (ABC/2) maintains inter- 11. Xue W, Vegunta S, Zwart CM, Aguilar MI, Patel AC, Hox-
nal consistency and allows for measurement of all scans. worth JM et al (2017) Retrospective validation of a computer-
assisted quantification model of intracerebral hemorrhage vol-
Acknowledgements Services in support of the research project were ume on accuracy, precision, and acquisition time, compared with
generated by the VCU Massey Cancer Center Biostatistics Shared standard ABC/2 manual volume calculation. Am J Neuroradiol
Resource, supported, in part, with funding from NIH-NCI Cancer 38(8):1536–1542. https://doi.org/10.3174/ajnr.A5256
Center Support Grant P30 CA016059. 12. Bland JM, Altman DG (1986) Statistical methods for assessing
agreement between two methods of clinical measurement. Lancet
Data availability The datasets generated during and/or analyzed dur- 1(8476):307–310
ing the current study are available from the corresponding author on 13. Lin LI (1989) A concordance correlation coefficient to evaluate
reasonable request. reproducibility. Biometrics 45(1):255–268
14. Carrasco JL, King TS, Chinchilli VM (2009) The concordance
correlation coefficient for repeated measures estimated by vari-
Compliance with ethical standards ance components. J Biopharm Stat 19(1):90–105. https://doi.
org/10.1080/10543400802527890
Conflict of interest The authors declare that they have no conflict of 15. Carrasco JL, Phillips BR, Puig-Martinez J, King TS, Chinchilli
interest. VM (2013) Estimation of the concordance correlation coeffi-
cient for repeated measures using SAS and R. Comput Methods
Ethical approval All procedures performed in studies involving human Programs Biomed 109(3):293–304. https://doi.org/10.1016/j.
participants were in accordance with the ethical standards of the insti- cmpb.2012.09.002
tutional and/or national research committee and with the 1964 Helsinki 16. Maeda AK, Aguiar LR, Martins C, Bichinho GL, Gariba MA
declaration and its later amendments or comparable ethical standards. (2013) Hematoma volumes of spontaneous intracerebral hemor-
This article does not contain any studies with animals performed by rhage: the ellipse (ABC/2) method yielded volumes smaller than
any of the authors. those measured using the planimetric method. Arq Neuropsiquiatr
71(8):540–544. https://doi.org/10.1590/0004-282X20130084
17. Sheth KN, Cushing TA, Wendell L et al (2010) Comparison
of hematoma shape and volume estimates in warfarin versus
non-warfarin-related intracerebral hemorrhage. Neurocrit Care
12(1):30–34. https://doi.org/10.1007/s12028-009-9296-7
References 18. Huttner HB, Steiner T, Hartmann M et al (2006) Comparison
of ABC/2 estimation technique to computer-assisted planimet-
1. Dolecek TA, Propp JM, Stroup NE, Kruchko C (2012) CBTRUS ric analysis in warfarin-related intracerebral parenchymal hem-
statistical report: primary brain and central nervous system orrhage. Stroke 37(2):404–408. https : //doi.org/10.1161/01.
tumors diagnosed in the united states in 2005–2009. Neuro STR.0000198806.67472.5c
Oncol 14(Suppl 5):v1–v49. https: //doi.org/10.1093/neuonc / 19. Divani AA, Majidi S, Luo X et al (2011) The ABCs of
nos218 accurate volumetric measurement of cerebral hematoma.
2. Krampla W, Newrkla S, Pfisterer W et al (2004) Frequency and Stroke 42(6):1569–1574. https : //doi.org/10.1161/STROK
risk factors for meningioma in clinically healthy 75-year-old EAHA.110.607861
patients: results of the transdanube ageing study (VITA). Cancer 20. Sreenivasan SA, Madhugiri VS, Sasidharan GM, Kumar RV
100(6):1208–1212. https://doi.org/10.1002/cncr.20088 (2016) Measuring glioma volumes: a comparison of linear
3. Vernooij MW, Ikram MA, Tanghe HL et al (2007) Incidental measurement based formulae with the manual image segmen-
findings on brain MRI in the general population. N Engl J Med tation technique. J Cancer Res Ther 12(1):161–168. https://doi.
357(18):1821–1828 org/10.4103/0973-1482.153999
4. Wiemels J, Wrensch M, Claus EB (2010) Epidemiology and 21. Dirks MS, Butman JA, Kim HJ et al (2012) Long-term natural his-
etiology of meningioma. J Neurooncol 99(3):307–314. https:// tory of neurofibromatosis type 2-associated intracranial tumors. J
doi.org/10.1007/s11060-010-0386-3 Neurosurg 117(1):109–117. https: //doi.org/10.3171/2012.3.JNS11
5. Chamoun R, Krisht KM, Couldwell WT (2011) Inciden- 1649
tal meningiomas. Neurosurg Focus 31(6):E19. https: //doi. 22. Kasuya H, Kubo O, Tanaka M, Amano K, Kato K, Hori T (2006)
org/10.3171/2011.9.FOCUS11220 Clinical and radiological features related to the growth poten-
6. Brem SS, Bierman PJ, Black P et al (2005) Central nervous tial of meningioma. Neurosurg Rev 29(4):293–296. https://doi.
system cancers: clinical practice guidelines in oncology. J Natl org/10.1007/s10143-006-0039-3. (discussion 296–7)
Compr Canc Netw 3:644–690 23. Kuratsu J, Kochi M, Ushio Y (2000) Incidence and clinical fea-
7. Hashiba T, Hashimoto N, Izumoto S et al (2009) Serial volumetric tures of asymptomatic meningiomas. J Neurosurg 92(5):766–770.
assessment of the natural history and growth pattern of inciden- https://doi.org/10.3171/jns.2000.92.5.0766
tally discovered meningiomas. J Neurosurg 110(4):675–684. https 24. Alva S, Eisenberg D, Duffy A, Roberts K, Israel G, Bell R
://doi.org/10.3171/2008.8.JNS08481 (2008) Virtual three-dimensional computed tomography assess-
8. Nakamura M, Roser F, Michel J, Jacobs C, Samii M (2003) ment of the gastric pouch following laparoscopic roux-Y gastric
The natural history of incidental meningiomas. Neurosurgery bypass. Obes Surg 18(4):364–366. https://doi.org/10.1007/s1169
53(1):62–70. (discussion 70-1) 5-008-9438-6
9. Chang V, Narang J, Schultz L et al (2012) Computer-aided volu- 25. Bendon CL, Sheerin FB, Wall SA, Johnson D (2014) The relation-
metric analysis as a sensitive tool for the management of inciden- ship between scaphocephaly at the skull vault and skull base in
tal meningiomas. Acta Neurochir (Wien) 154(4):589–597. https sagittal synostosis. J Craniomaxillofac Surg 42(3):245–249. https
://doi.org/10.1007/s00701-012-1273-9. (discussion 597) ://doi.org/10.1016/j.jcms.2013.05.009
13
Journal of Neuro-Oncology
26. Davies RS, Abdelhamid M, Vohra RK, Bradbury AW, Adam measurements for le fort I osteotomy. Int J Oral Maxillofac Surg
DJ (2012) The relationship between aortic aneurysm sac throm- 43(2):197–203. https://doi.org/10.1016/j.ijom.2013.06.005
bus volume on coagulation, fibrinolysis and platelet activity. 33. Perkins JA, Sidhu M, Manning SC, Ghioni V, Sze R (2005) Three-
Thromb Res 130(3):463–466. https://doi.org/10.1016/j.throm dimensional CT angiography imaging of vascular tumors of the
res.2012.03.018 head and neck. Int J Pediatr Otorhinolaryngol 69(3):319–325
27. Wyss TR, Dick F, England A, Brown LC, Rodway AD, Green- 34. Greess H, Nomayr A, Tomandl B et al (2000) 2D and 3D visuali-
halgh RM (2009) Three-dimensional imaging core laboratory sation of head and neck tumours from spiral-CT data. Eur J Radiol
of the endovascular aneurysm repair trials: validation of meth- 33(3):170–177
odology. Eur J Vasc Endovasc Surg 38(6):724–731. https://doi. 35. Kwon J, Barrera JE, Jung TY, Most SP (2009) Measurements of
org/10.1016/j.ejvs.2009.09.007 orbital volume change using computed tomography in isolated
28. Yeung KK, van der Laan MJ, Wever JJ, van Waes PF, Blanken- orbital blowout fractures. Arch Facial Plast Surg 11(6):395–398.
steijn JD (2003) New post-imaging software provides fast and https://doi.org/10.1001/archfacial.2009.77
accurate volume data from CTA surveillance after endovascular 36. Richard MJ, Morris C, Deen BF, Gray L, Woodward JA (2009)
aneurysm repair. J Endovasc Ther 10(5):887–893. https://doi. Analysis of the anatomic changes of the aging facial skeleton
org/10.1177/152660280301000507 using computer-assisted tomography. Ophthal Plast Reconstr Surg
29. Day NJ, Earnshaw D, Salazar-Ferrer P, Walsh CJ (2013) Preop- 25(5):382–386. https://doi.org/10.1097/IOP.0b013e3181b2f766
erative mapping of fistula-in-ano: a new three-dimensional MRI- 37. Wang H, Li W, He H, Luo L, Chen C, Guo Y (2013) 320-detec-
based modelling technique. Colorectal Dis 15(11):e699–e701. tor row CT angiography for detection and evaluation of intrac-
https://doi.org/10.1111/codi.12438 ranial aneurysms: comparison with conventional digital sub-
30. de Jonge GJ, van Ooijen PM, Overbosch J, Gueorguieva AL, traction angiography. Clin Radiol 68(1):e15–e20. https://doi.
Janssen-van der Weide MC, Oudkerk M. Comparison of (semi-) org/10.1016/j.crad.2012.09.001
automatic and manually adjusted measurements of left ventricular 38. Wilson DO, Ryan A, Fuhrman C et al (2012) Doubling times and
function in dual source computed tomography using three differ- CT screen-detected lung cancers in the pittsburgh lung screen-
ent software tools. Int J Cardiovasc Imaging 2011;27(6):787-794. ing study. Am J Respir Crit Care Med 185(1):85–89. https://doi.
https://doi.org/10.1007/s10554-010-9727-8 org/10.1164/rccm.201107-1223OC
31. Gaia BF, Pinheiro LR, Umetsubo OS, Costa FF, Cavalcanti MG 39. Shah GD, Kesari S, Xu R et al (2006) Comparison of linear and
(2013) Comparison of precision and accuracy of linear measure- volumetric criteria in assessing tumor response in adult high-
ments performed by two different imaging software programs and grade gliomas. Neuro Oncol 8(1):38–46
obtained from 3D-CBCT images for le fort I osteotomy. Den- 40. Greenberg MS. Handbook of neurosurgery, 8th edn. New York:
tomaxillofac Radiol 42(5):20120178. https://doi.org/10.1259/ Thieme. c2016. Chapter 42, Meningiomas; pp 690–700.
dmfr.20120178
32. Gaia BF, Pinheiro LR, Umetsubo OS, Costa FF, Cavalcanti MG Publisher’s Note Springer Nature remains neutral with regard to
(2014) Validity of three-dimensional computed tomography jurisdictional claims in published maps and institutional affiliations.
13