Single Voxel MR Spectroscopy in Non-Herniated Painful, Herniated Painful, and Non-Painful Lumbar Discs

L Tanenbaum1, S·Hu2, M·Gornet3, S·Berven2, J·Peacock4, J·Lotz2,4, D·Bradford4,·J Claude4, P·Kane4, F·Schranck3, D·Stewart5

1Icahn School of Medicine at Mount Sinai, 2University of California San Francisco, 3Orthopedic Center of Saint Louis, 4Nocimed, LLC,

5MRI Partners of Chesterfield1

PG

Lactate

col

Disclosures

• Tanenbaum, L: Nocimed, LLC - advisory board• Bradford, D; Lotz, J: Nocimed, LLC - management,

shareholder, royalty• Kane, P; Claude, J: Nocimed, LLC - shareholder,

consulting• Peacock, J: Nocimed, LLC - management, shareholder• Schranck, F; Gornet, M: The Orthopedic Center of St.

Louis - Nocimed sponsored research grant, Nocimed, LLC - shareholder

2

Purpose

3

• Evaluate non-invasive in vivo single voxel MR spectroscopy (SVS) in lumbar discs on two commercial 3T scanners - GE Signa & Siemens Verio for

a) discogenic low back pain (LBP) pain patients receiving provocative discography (PD), and

b) asymptomatic control volunteers (ASY) • Correlate results to clinical phenotypes– non-herniated painful (NHP)– herniated painful (HP)– non-painful (NP) discs

BackgroundDisc degeneration

• altered disc structure and biochemical composition that may or may not cause pain– decreased proteoglycan synthesis– increase matrix catabolism

• features that differentiate painful from non painful degenerative discs not well understood.

*Buser, 2010 Spine: Affiliated Society Meeting Abstracts

• consistent changes in markers of hypoxia, innervation and inflammation showed strong correlation with discs associated with pain

*Buser, 2010 Spine: Affiliated Society Meeting AbstractsNucleus samples

Gene expression with highest correlation between painful vs. non-painful discs relate to (R2=.67):

1. hypoxia - lactate dehydrogenase; 2. innervation - nerve growth factor.

*Buser, 2010 Spine: Affiliated Society Meeting Abstracts

BackgroundDisc degeneration

• Painful discs are distinguished by hypoxia, inflammation, neovascularization & neoinnervation*

• Consistent changes in biomarkers/chemical mediators detectable and quantifiable with MR spectroscopy

*Freemont, A. J., (2009) Rheumatology; 48:5–10.

BackgroundIdentifying the painful disc

Background-Ex vivo Disc SVS, 11T HR MAS Surgical Samples – PG/LA

PG

Lactate

col

Painful Disc(PD+, LBP)n=9

PG

Lactate

col

Non-Painful Disc(Scoliosis)n=9

8

Keshari, Spine 2008. 33(3): p. 312-317

p<.005

Materials & Methods

• Iterative technical development protocol– Conducted in series: GE (2008-10), Siemens (2011-13)– T1/T2 weighted imaging sequences– SVS acquisition, each lumbar disc sequentially• Inter-platform differences:

– OVS bands: GE only (not used for Siemens)– Autoshim Auto + Manual shim: Siemens protocol refinement

– Manual voxel localization in disc to remove end-plates

9

Materials & Methods• Acquisition Systems – GE Signa HDxt 3.0T, v14x

• HDCTL 456 spine coil• Pulse Sequence: custom Chess/Press

– “modified” PROSETM for disc SVS: TR=1500, TE=28ms

– Siemens 3T Verio, Model 10276755 Numaris/4 • Spine Coil Model 8622743• Pulse Sequence: SVS Syngo VB17 Version 1.0

– investigational “patch” for 3-5mm voxel height

• Both single center, Nocimed sponsored, IRB approved– GE 3T: University of California at San Francisco– Siemens 3T: The Orthopedic Center of St. Louis

10

Non-Painful Disc

Non-Painful

Materials & Methodssignal post-processing

Automated Channel Select Frame Edit Phase/Freq Apodize Aggregate Baseline

11

Materials & Methodstechnical “success” & signal quality assessment

12

• GE data: initial in vivo development cycle of entire approach• Reported data not include technical failures for poor signal quality• Technical “failures” = only “indeterminate” results of Dx correlation

algorithm

• SIEMENS data: x-platform application/refinement of prior learning • Less technical development required• Reported data includes all completed acquisitions - signal quality

successes & failures• Primary Signal (satisfactory) Quality Assessment Criteria:o SNR > 3 for PG or LA peak regionso Water FWHM < 35 Hzo Presence (fail) or Absence (accept) of lipid motion artifact

• Different approach per platform, serial development phases• Earlier GE platform data:

– Multi-variate (PG & LA) automated statistical algorithm development via “JMP” logistic regression software

– Predictive value of SVS biomarker measures for painful vs. non-painful

• Later Siemens platform data:– Simple PG/LA ratio– Predictive value/statistical modeling planned

(> more pain patients)

13

Materials & Methodsdiagnostic interpretation/correlation to disc type

1%

99%

1%

14

Materials & Methodsstudy design – inclusion criteria

• Male and non-pregnant females, 18-70 yrs old • Pain group:

– Oswestry Disability Index ≥ 40%, or Visual Analog Scale ≥ 4– Indicated for Lumbar Spine Provocation Discography (PD)

• Practice Guidelines for Spinal Diagnostic and Treatment Procedures (ISIS, 2004)

• PD performed 6wk-6mo before, or 1mo after, Disc SVS exam– Up to 12 mo before in earlier GE study

• Asymptomatic control volunteers:– Absent chronic severe LBP– VAS≤1, ODI≤10.

15

Materials & Methodsstudy design – exclusion criteria

• Complicating extradiscal pain sources– fracture, spondylolisthesis, spondylolysis

• Prior lumbar surgery or intradiscal treatments– except diagnostic provocative or anesthetic

discography• Chronic disease or pain, or psychological

dysfunction, relevant to study objectives

GE studyn= 42 Subjects, 79 Discs1

16

Pain Pts (PD)n=14

Volunteersn=28

Painful Discs(PD+)

Non-Painful Discs(PD-)

Non-Painful Discs(assumed)

n = 15 n = 49

n = 15+ Control, Painful

n = 64- Control, Non-Painful

Siemens studyn=33 Subjects, 130 Discs1

17

Pain Pts (PD)n=14 Volunteers

n=19

Non-HerniatedPainful Discs

(PD+)

HerniatedPainful Discs

(PD+)

Non-Painful Discs

(PD- & remote levels)

Non-Painful Discs

(assumed)

n = 32 n = 9

n = 16 n = 1031

n = 1191

1 Includes n=1 disc excluded from analysis as non-lumbar (S1-S2)2 Includes n=1 disc analyzed separately due to “extra”discal co-morbidities (spondylosis + facet arthropathy)

18

+

-

Results- 3T GE Signa Case 1, Pain Pt (PD+ L5S1, PD- L4L5)

Discogram: +/-

Hu, Orthopaedic Research Society 2010

19

Results - GE 3T Signa 94% (74/79) Determinate, 99% (73/74) Accurate (PG,LA)

Hu, Orthopaedic Research Society 2010

SVS Positive n=13 all true+*

SVS Negative n=60 true-, 1 false-*

SVS Indeterminate n=5

PG &

LA-

rela

ted

Dx

Algo

rithm

“sc

ore”

Case 2 – 3T Siemens VerioNHP Pain Pt, 2-level PD+/-

L5-S16mm

L4-L56mm

L3-L45mm

L2-L35mm

L1-L24mm

10

0

22

Discogram: +/-

Results - Siemens 3T VerioSignal Quality Success 85% (100% Manual Shim, >1mm)

23

Total Autoshim only Auto+Manual Shim0

20

40

60

80

100

120

140

110

74

36

7

6

1

11

10

1

2

2

0

Lipid/motionLow SNR < 3xPoor Shim (WaterFWHM > 35Hz)Success

Technical Success, Signal Quality - Siemens 3T Verio Overall vs. Autoshim only vs. Autoshim + Manual Shim (Later Protocol Rev)

Dis

cs (

#)

Protocol Development (Shim)

Both 1mm Voxels

20%

80%

95%

5%

92 Discs21 Sub-

jects

38 Discs11 Subjects

15%

85%

130 Discs32 Subjects

24

L2L34mm

L1L24mm

Autoshim Only(Fail Water FWHM)

Auto+Manual shim

N313

Asymptomatic Control Volunteer

Results - Siemens 3T VerioAutoshim only vs. Auto+Manual Shim – Example

Conclusionsdisc SVS signal quality - reliability

• 100% robust disc spectra per signal quality criteria @ ≥3mm disc height per later protocol refinements (autoshim + manual shim) of the Siemens 3T platform study

• 80% reliability w/ autoshim only

25

Results – Siemens 3T VerioPG/LA Dx Correlation vs. Disc “Type” Controls (n=110)

26

0 1 2 3 4 50.20

2.00

20.00

Disc SVS PG/LA Ratio vs. Disc Type

log

PG/L

A Ra

tio ,

Lum

bar D

isc S

VS

PD+Non-herniated

PD+Non-herniated,Spondy

PD+Herniated

NC(PD- and ASY)

n=1n=1 n=100n=8

27

Painful Non-Painful0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

93% 100%

7%0%

Mis-Match vs. Control

Match vs. Control

Dx Correlation Results, GE & Siemens 3T Combined(PG & LA-based Dx Algorithms separately developed, MR platform-specific)

Dis

cs (

#)

n=13

n=1

n=161

n=0

Control Disc Type

Combined Siemens & GE 3T ResultsPG & LA correlate 174/175 (>99%) discs, 72 subjects

Conclusions

• Combined PG & LA-related results from the two serial GE & Siemens 3T platform disc SVS development studies reflects >99% correlation between classic “contained” discogenic pain and non-painful discs in 175 total discs in 72 subjects

– 100% PPV for pain in 13 painful PD+ discs

– 0% False+ in 161 non-painful discs

28

Conclusions

29

• Disc SVS may be most appropriate for classic contained axial discogenic low back pain w/o extradiscal pain generators, vs. pain caused by extradiscal morbidities (e.g. herniations, facet, vertebral abnormalities)–may still enhance Dx regimen in multi-factorial cases

involving both intra-discal and extra-discal pain sources

Single Voxel MR Spectroscopy in Non-Herniated Painful, Herniated Painful, and Non-Painful Lumbar Discs

L Tanenbaum1, S·Hu2, M·Gornet3, S·Berven2, J·Peacock4, J·Lotz2,4, D·Bradford4,·J Claude4, P·Kane4, F·Schranck3, D·Stewart5

1Icahn School of Medicine at Mount Sinai, 2University of California San Francisco, 3Orthopedic Center of Saint Louis, 4Nocimed, LLC,

5MRI Partners of Chesterfield30

PG

Lactate

col