clinical application MRS

CLINICAL APPLICATION OF PROTON MAGNETIC RESONACE SPECTROSCOPY 1H-MRS IN THE DIAGNOSIS OF INTRA-AXIAL BRAIN TUMOR IN ADULTS Cao Thiên Tượng Phạm Ngọc Hoa...  MRS can help to differenti

CLINICAL APPLICATION OF

PROTON MAGNETIC RESONACE SPECTROSCOPY (1H-MRS) IN THE DIAGNOSIS

OF INTRA-AXIAL BRAIN

TUMOR IN ADULTS

Cao Thiên Tượng Phạm Ngọc Hoa

 Diagnosis of brain tumors based on CT and conventional MRI is still challenging problem

 Magnetic resonance spectroscopy (MRS) is

advanced technique being used to diagnosis brain tumors and other neurologic diseases

in many countries

 This technique provides information related

to cell membrane proliferation, neuronal

damage, energy metabolism and necrotic

transformation of brain or tumor tissues

 MRS can help to differentiate between

neoplastic and non-neoplastic lesions,

evaluate tumor tissue and edema, predict

grading of glioma, localize biopsy site,

evaluate after treatment (surgery,

radiotherapy…)

Metabolites in the brain in proton spectroscopy at 135ms TE, 1.5T (*)

Abbreviation Metabolite Frequency Significance

Cho Phosphocholine,

Glycerophosphorylcholine

3.22 ppm Membrane turnover, cell

proliferation

Cr Creatine and phosphocreatine 3.03 and

4ppm Temporal store for energy-rich

phosphates NAA N-acetyl-L-aspartate 2.01ppm Indicates the presence of

intact glioneural structures.

Exact function unknown

(inverte d)

doublet

Anaerobic glycolysis

Lipids Free fatty acide 1.2-1.4 ppm Necrosis

1H-MRS in the brain of normal

indiviual

Resonance Signal

intensity

Summary of MRS findings for brain lesions

(*)

High-grade glioma Very high Very high Very low

Gliomatosis cerebri Normal/High Normal/High Low

Cho/Cho(n): Cho lesion/cho contralateral

 Evaluate the clinical application of

proton magnetic resonance

Material and methods (1)

neurosurgery of Cho Ray hospital with

supratetorial intra-axial brain lesion suspiciuos for a neoplasm on clinical examination and

conventional MR imaging Those patients had confirmed on pathologic examination (n=26)

examination and follow-up after treatment in the department of neurology of Cho Ray

hospital (n=3)

 Prospective study

 Study duration from Jun 2008 to Oct 2008

 Patients are examined by conventional MR

imaging on 1T MRI unit (Harmony, Siemens, Erlangen, Germany) or 1.5 Tesla unit (Avanto, Siemens, Erlangen, Germany)

 Multivoxel 1H-MRS examination is performed

on 1.5 Tesla MRI unit (Avanto, Siemens,

Erlangen, Germany) with a ̣2D-Chemical shift imaging technique [TR/TE: 1500/135ms]

Material and methods (2)

 Automated postprocessing was performed with

NUMARIS 4 software package, version Syngo MR B

15 (Siemens)

 Cho (3.22ppm), Cr (3.02ppm), NAA (2.02ppm)

were identified Lactate peak (1.3ppm) was noted

by qualitative analysis.

 Calculate ratios: Lesion Cho / contralateral side

Cho (Abv nCho), lesion NAA / contralateral side

NAA (abv nNAA), Lesion Cho /lesion Cr (abv

lChocr), lesion Cho /lesion NAA (Abv lChoNAA),

Lesion NAA /lesion Cr (abv lNAACr), Lesion Cho/ contralateral side Cr and lesion NAA /contralateral side Cr

 Statistical analysis by SPSS software 10.0.5 (SPSS

Material and methods (3)

+20 gliomas were graded according to

WHO classification from Grade I to Grade

IV (grade I, n=1; grade II, n=8, grade III, n=8, grade IV, n=3), including 19

astrocytomas and 1 oligodendroglioma +2 brain metastase.

+1 CNS lymphoma.

+3 brain abscesses.

General sex distribution

37.5%

62.5%

nu

nam

Sex distribution of brain tumor

group

Distribution of groups Tumor (n=23), non-tumor (n=6), Normal

Age mean of groups

Metabolite Ratios in tumor group and non-tumor group

Tumor N=23

(*) Mann-Whitney U test, P<0.05

Cho (nCho) Ratio in tumor group

and non-tumor group

6 23

N =

loai ton thuong

khong u u

6 23

N =

loai ton thuong

khong u u

26

NAA (nNAA) Ratio in tumor group and non-tumor group

Cho/Cr Ratio in tumor group,

non-tumor group and normal

individuals

11 6

23

N =

loai ton thuong

binh thuong khong u

22

11 6

23

N =

loai ton thuong

binh thuong khong u

11 6

23

N =

binh thuong khong u

NAA/Cr Ratio in tumor group, non-tumor group and normal

individuals

ROC curve predict tumor/non-tumor lesion in nCho

ratio

Sensitivity 91.3%, Specificity 100%, PPV 100%, NPV 75%, p=0.0001, (95% CI: 0.848-0.989) cutoff nCho>1.32 AUC: 0.982

Sensitivity 78.3%, Specificity 83.3%, PPV 94.7%, NPV 50%, p=0.0007, (95% CI: 0.670-0.955) cutoff nNAA</=0.3 AUC:

ROC curve predict tumor/non-tumor lesion in nNAA

ratio

Sensitivity 95.65%, Specificity 100%, PPV 100%, NPV 85.7%, p=0.0001, (95% CI: 0.866-1.000) cutoff: Cho/Cr>1.8 AUC: 0.993

ROC curve predict tumor/non-tumor lesion in Cho/Cr

ratio

Sensitivity 100%, Specificity 100%, PPV 100%, NPV 100%, p=0.000, (95% CI: 0.879-1.000) cutoff: Cho/Cr>1.33 AUC: 1.000

ROC curve predict

tumor/non-tumor lesion in Cho/NAA ratio

Comparison between low-grade gliomas (I-II) and high-grade gliomas (III-IV) in metabolite

ratios

Low-grade N=9 High-gradeN=11 P Mean (SD) Medium (min-max) Mean (SD) Medium (min-max)

nCho 2.65 (1.64) 2.26 (1.16-6.44) 2.25 (0.85) 2.33(1.28-3.83) 0.941 *

nNAA 0.33 (0.21) 0.27 (0.07-0.74) 0.15 0.11(0.05-0.32) 0.038

lCho/NAA 7.68 (5.77) 6.67(2.00-19.93) 11.30(6.83) 10.0(2.78-21.29) 0.261 * lNAA/Cr 0.54 (0.35) 0.53 (0.11-1.26) 0.45 (0.26) 0.42(0.15-1.1) 0.503 * lCho/Cr 3.08 (2.42) 2.28 (1.71-9.35) 4.07 (2.00) 3.42(1.81-7.98) 0.131 * lCho/colCr 2.42 (1.59) 1.73 (1.12-6.07) 2.02 (0.99) 1.84 (0.76-4.22) 1.000 * lNAA/colCr 0.43 (2.82) 0.38 (0.11-1.04) 0.23 (0.17) 0.21 (0.10-0.71) 0.056 *

Cho/NAA Ratio in low-grade (I-II) and high-grade

(III-IV)

11 9

20

N =

grade

cao thap

11 9

20

N =

grade

cao thap

and high-grade (III-IV)

Mean ratio of Cho/Cr, Cho/NAA and

NAA/Cr in gradeI-II, III and IV

Low-grade (I-II)

N=9 Grade IIIN=8 P Mean (SD) Medium (min-

max) Mean (SD) Medium (min-max)nCho 2.65 (1.64) 2.26 (1.16-6.44) 2.28 (0.85) 2.34(1.28-3.83) 0.963 *

nNAA 0.33 (0.21) 0.27 (0.07-0.74) 0.10 0.10(0.05-0.18) 0.008

lNAA/Cr 0.54 (0.35) 0.53 (0.11-1.26) 0.45 (0.26) 0.42(0.15-1.1) 0.139 * lCho/Cr 3.08 (2.42) 2.28 (1.71-9.35) 4.52 (2.18) 4.80(1.81-7.98) 0.139 * lCho/colCr 2.42 (1.59) 1.73 (1.12-6.07) 2.14 (1.07) 1.99 (0.98-4.22) 0.963 *

Comparison between low-grade gliomas (I-II) and grade III gliomas in metabolite ratios

ROC curve predict low/high grade of

glioma in nNAA ratio

Sensitivity 63.6%, Specificity 88.9%, PPV 87.5%, NPV 66.7%, p=0.011, (95% CI: 0.539-0.929) cutoff NAA</=0.86

Sensitivity 72.7%, Specificity 77.8%, PPV 80%, NPV 70%, p=0.08, (95% CI: 0.464-0.885) Cutoff Cho/Cr > 2.86 AUC:

ROC curve predict low/high grade of glioma in Cho/Cr ratio

Sensitivity 90.9%, Specificity 66.7%, PPV 76.9%, NPV 85.7%, p=0.0308, (95% CI: 0.506-0.911) Cut off </= 0.32 AUC: 0.747.

ROC curve predict low/high grade of glioma in lesion NAA/contralateral side

Cr ratio

Qualitative evaluation of lactate

peak

(44%) tumor cases and 1/6 (17%)

non-tumor case All normal cases is absence of lactate peak.

cases of low-grade gliomas and 7/11 cases (64%) of high-grade tumors.

A 26-year-old man with CNS

A 49-year-old man with anaplastic astrocytoma grade

Normal, contralateral

A old man with brain metastasis

64-year-Normal, contralateral

Normal, contralateral

Lac

Discussion (1)

 Our result indicates elevated Cho (nCho) in

tumor cases (mean 2.39) compared to

non-neoplastic lesions (mean 0.88) In the ROC

curve, a nCho > 1.32, predicted tumor with sensitivity 91.3%, Specificity 100%, PPV

100%, NPV 75%

 Study of Nagar VA et al (*) showed nCho> 1

in tumor lesions and nCho< 1 in non-tumor

lesions Several other studies have shown

similar results

 This result indicates Cho could be used as an MRS biomarker in predicting of brain tumor

(*) Nagar, V.A., et al., Multivoxel MR spectroscopic imaging distinguishing

intracranial tumours from non-neoplastic disease Ann Acad Med Singapore,

 nNAA significantly decreases in tumor case (mean: 0.24) compared to non-tumor cases (mean: 0.52)

decrease or displacement of neurons by

tumor (*)

non-neoplastic diseases as infarction,

decrease in NAA is also indicator to suggest brain tumor

Discussion (2)

Result of Hourani & al.(*) and our study

Hourani R& al.

N=36 Our studyN=23 Hourani R& al.N=33 Our studyN=6

(*) Hourani, R., et al., Can proton MR spectroscopic and perfusion imaging

differentiate between neoplastic and nonneoplastic brain lesions in adults?

predicting brain tumor

 Our result indicates significant differences in Cho/NAA, Cho/Cr và NAA/Cr between tumor group, non-tumor and normal individuals

This result is consistency with many other

studies

 Based on the analysis of ROC curve in the

metabolite ratios, we find that Cho/NAA has most probability of predicting brain tumor

Discussion (4)

predict tumor with a cutoff value >

1.64, accuracy was 84.1%, sensitivity 86,1%, specificity 81,8%, PPV 83.8%, NPV 84.4% (*)

 Our result has higher sensitivity,

specificity due to small sample size

of non-neoplasitc lesions and only

consist of abscess and infarct cases

(*) Hourani, R., et al., Can proton MR spectroscopic and perfusion imaging

differentiate between neoplastic and nonneoplastic brain lesions in adults?

Discussion (5)

of MRS for predicting grade of glioma Generally, those studies have shown that 1H-MRS could be useful in

grading gliomas

MRS, DWI and PWI increases the

diagnostic accuracy in grading

gliomas.

Discussion (6)

histopathologic bias in grading

gliomas, our result has limited value

to predict malignancy

difference between low-grade glioma and high-grade glioma in nNAA There was no statistically significant

diference between high-and low-grade tumors in other ratios

Discussion (7)

 Our result shows significant difference

between grade I-II glioma and grade III glioma in nNAA, Cho/NAA and

NAA/contralateral Cr That indicates that, Cho level of grade IV gliomas in our study

is not too high, there may be necrosis within tumor.

 Several studies have been found that Cho

may be reduced because of the presence

of necrosis in GBM (*)

Discussion (8)

 Although there was no statistically significant

difference between neoplastic and neoplastic lesions, high-and low grade glioma in lactate peak, there is more presence of Lac peak than in tumor cases and almost in high-grade tumors

non- Lipid and lactate elevation correlate with

necrosis in high-grade gliomas and have also been shown to be useful in differentiating

between low and high-grade gliomas (*)

(*) D.P Soares, M.L., Magnetic resonance spectroscopy of the brain:

Conclusion

brain tumor.

imaging studies helps more

differentiation between brain

neoplasitic-and non-neoplastic lesions.

malignancy of glioma.