Design and Performance of a Localized Fiber Optic, Spectroscopic Prototype Device for the Detection of the Metabolic Status of

1
Design and Performance of a Localized Fiber
Optic, Spectroscopic Prototype Device for the
Detection of the Metabolic Status of Vulnerable
Plaquein-vitro Investigation of Human Carotid
Plaque
2
OUTLINE

• INTRODUCTION
• Problem identification, objectives, specific
  aims, hypotheses, and background review
• OPTICAL DESIGN
• 1 mm3 tissue volume interrogation achieved with
  optical probe
• METHODS
• Laboratory setup
• Data collection
• Calibration model development
• RESULTS
• DISCUSSION/CONCLUSIONS
• Limitations
• Future work

3
PROBLEM IDENTIFICATION

• Atherosclerotic cardiovascular disease 6.3 M
  deaths / yr worldwide
• Cardiovascular disease 1 killer in the U.S.
• 1.5 M myocardial infarctions (MI) / yr in the
  U.S.
• 250,000 / yr sudden cardiac deaths
• 111.8 billion / yr health care costs
  (direct/indirect)
• Major risk factors
• Smoking
• High blood cholesterol (LDL/HDL ratio)
• Physical inactivity
• Overweight/Obesity
• Diabetes mellitus

Source American Heart Association. 2002 Heart
and Stroke Statistical Update. 2001.
http//www.americanheart.org
4
Sudden Cardiac Death
Acute MI
Vulnerable Plaque(s)
Everybody has atherosclerosis, the question is
who has vulnerable plaque
5
Unknown Diagnosis Vulnerable Plaque

• The precursor that ultimately ends in acute
  thrombi (clots) of sudden death MI
• Inflammatory cells found preferentially in
  vulnerable plaque
• Activity sustained through anaerobic metabolism
  and lactate production

6
Morphology vs. Activity Imaging
Thermography, Spectroscopy, immunoscintigraphy,
MRI with targeted contrast media
Show Different Activity
Inactive and non-inflamed plaque
Active and inflamed plaque
Appear Similar in
Morphology
IVUS
MRI w/o CM
OCT
7
HISTOLOGY
THROMBUS
LIPID CORE
J Am Coll Cardiol. 2001 Sep38(3)718-23.
Am J Pathol. 2000 Oct157(4)1259-68.
FIBRO-CALCIFIC
Courtesy of Texas Heart Institute
8
LONG-TERM OBJECTIVES

• Develop an optical spectroscopy catheter system
  to determine the metabolic status of
  atherosclerotic vessels
• No exogenous dyes
• No ionizing radiation
• Low cost addition to existing cardiac
  catheterization laboratory
• Locate and identify vulnerable plaque based on
  metabolic status with optical spectroscopy

9
SHORT-TERM GOALS

• Demonstrate feasibility in-vitro of optical
  spectroscopy to accurately determine metabolic
  status
• Tissue lactate concentration
• Tissue pH

10
(No Transcript)
11
SPECIFIC AIMS

1. Design a reflectance-based fiber optic probe that
   uses visible to near-infrared light optimally to
   interrogate a small volume of tissue.
2. Estimate the depth penetration of fiber optic
   probe, based on theory and experiments.
3. Identify major interferents to the optical
   spectra and tissue reference measurements
   collection.
4. Collect and analyze fresh tissue from human
   carotid endarterectomies to create large optical
   calibration training set while maintaining tissue
   in a viable physiological state in-vitro.

12
HYPOTHESES

1. A small fiber optic prototype can make optical
   measurements in-vitro for the assessment of
   metabolically active plaque in a defined region
   of tissue (lt 1 mm3 volume).
2. in-vitro experimental factors can be assessed to
   their importance in affecting the optical
   calibration accuracy. The tissue temperature,
   experiment time, and gross pathology are
   identified a priori.
3. Mathematical models can be developed which relate
   the corresponding optical spectra to the
   individually measured metabolic parameters
   (tissue pH and lactate concentration) in the
   presence of inherent pathological variability.

13
SPECTROSCOPY BASICS
In general, spectroscopy is the use of the
electromagnetic spectrum to perform physical or
chemical analysis
14
PREVIOUS WORK

• Optical spectroscopy proposed by Lodder (UKY),
  Feld (MIT) and Jaross (Germany) to characterize
  morphological properties of atherosclerotic
  plaques such as thin fibrous cap, large lipid core

15
LACTATE AND PLAQUE

• Metabolite produced by activated macrophages
• Studies show lactate is present in plaque (Kirk,
  Zemplenyi)
• Anaerobic glycolysis LDH
• Pyruvate NADH ? Lactate NAD
• Overall anaerobic process
• Glucose 2ADP 2Pi ? 2 Lactate 2ATP 2H20
  2H

16
NIR Spectrum
Near infrared absorbance of lactic acid
17
PLAQUE pH
Large scale, ex-vivo study on carotid plaques
demonstrated metabolic heterogeneity in grossly
pathological areas (Grascu, 1999)
Inflamed regions of plaque are lower in pH in the
atherosclerotic Watanabe rabbit and human carotid
plaques plaque pH heterogeneity demonstrated
(Naghavi, 2002)
18
DR. SOLLERS LAB

• Tissue pH can be measured by NIR spectroscopy in
  heart muscle (Soller, Zhang 1998)
• Lessons learned volume of optical measurement gtgt
  volume of reference measurement
• Heterogeneity in a large tissue volume may be
  solved with smaller optical probe

19
OPTICAL DESIGN
20
DESIGN PROCESS

• Define of optical probe requirements
• Theoretical considerations of tissue optical
  properties
• Monte Carlo simulations interpretation
• Building and testing several optical probes
• Depth penetration assessment

21
Optical Catheter System Diagram

• Optical fibers carry light to tissue
• Light is reflected and/or backscattered toward
  fibers that return light to spectrometer and
  tissue absorbance calculated
• Catheter geometry and optical coupling important
• Small source-receiver separations light
  penetrates tissue while restricting volume
  interrogated

tissue interface
2 mm
Absorbance
To spectrometer
Light in
wavelength
22
THEORY

• Monte Carlo Simulations
• Estimate light paths in complex absorbing and
  scattering medium
• Random events reflection, absorption,
  scattering, or transmission
• Define grid geometry, specify tissue optical
  properties

23
Diffuse Reflectance(radius)
normal
atherosclerotic
Theoretical Depth Penetration
24
OPTICAL EXPERIMENTS

• Compared signal-to-noise ratios (SNR) for several
  fiber types / configurations
• Different core sizes / number of fibers
• With or without optical windows
• Source-receiver separations

25
PROBE GEOMETRY
Large OD Probe
360 degree illumination w/ optical window
Forward illumination No optical window
Final probe W/ optical window
26
TISSUE PENETRATION STUDY

1. Reference spectra collected for each optical
   configuration (50, 500 micron separations).
2. Absorbance spectra collected with n-th slice of
   50 micron tissue.
3. Second absorbance spectrum collected with n-th
   slice plus diffuse reflector. Both absorption and
   scattering attenuate tissue signal.

27
(No Transcript)
28
FINAL PROBE DESIGN

• Using a source-receiver separation of 50 microns,
  adequate depth resolution could be achieved in
  plaque in both the visible and near-infrared
• Increasing the collection fiber core diameter
  size to 200 microns with improved transmission
  out to 2400 nm, higher signal-to-noise ratio is
  achieved by improving the fiber collection area
  by 4 times and collection efficiency
• Using a 0.5 mm thick quartz optical window fused
  on the common end, with forward-viewing optics,
  the signal-to-noise ratio would be further
  improved across all wavelengths

29
1 cm
Fiber optic probe used for all optical
determinations in this study.
30
METHODS
31
Laboratory setup for all studies.
32
Humidified Incubator maintained at 37C.
33
in-vitro Plaque Validation Study

• Minimum Eagles Medium (MEM), pH 7.4, 5.6 mM
  glucose, 26.2 mM NaHCO3, with non-essential amino
  acids was used (Invitrogen).
• Media equilibrated with 75 O2 / 5 CO2 gas
  mixture prior to plaque addition.
• Seven human carotid plaques (UMass Memorial IRB
  Approval 10041) were collected and placed
  immediately in 37C media enclosed in a
  humidified incubator at 37C.
• Two plaques that were not placed in the liquid
  media, only in the humidified air of the
  incubator, served as controls.
• Measurements were taken with a 0.5 mm OD
  multi-parameter sensor placed in the tissue
  (Diametrics, MN).
• Changes in tissue pH, temperature, PO2 and PCO2
  over time were analyzed.

34
Control plaque with multi-parameter sensor in
place.
35
Box-whisker plots for ?pH / hour and ?temperature
/ hour (top and bottom, respectively) for the
control and test plaques. The change in pH and
temperature over time is greater in the controls
than the test plaques.
36
STABILITY REVIEW

• Experiment time fixed max. 4 hrs
• in-vitro experimental stability criteria met
• lt0.03 pH units/hr and lt 0.4C /hr
• Tissue temperature in media gt 32C to ensure
  tissue viability
• Tissue values stable and different from media,
  controls
• Plaques in oxygenated media had higher pO2
  readings versus control plaques
• Thickness of plaque affected magnitude of pO2
  readings
• Unable to measure calcified areas over time

37
OPTICAL CALIBRATION

• 24 additional human carotid plaques were
  collected and placed in-vitro.
• Absorbance spectra (667 2500 nm) of each area
  were taken using Nicolet FTIR 670 spectrometer
  with fiber optic probe (Remspec, MA) for optical
  lactate determination.
• Tissue biopsies of the same area were taken using
  a 4-mm punch biopsy and immediately frozen in
  liquid nitrogen.
• Reference tissue lactate (LA) assayed using
  micro-enzymatic methods. Values reported as
  micromole LA per gram wet tissue.
• Matching spectra and reference values modeled by
  multivariate calibration techniques. R2 and the
  standard error of cross-validation (SECV) used to
  assess model accuracy.

38
OPTICAL CALIBRATION

• Absorbance spectra (400-1100 nm) were collected
  for a smaller subset of 14 plaques using a
  Control Development spectrometer and same optical
  probe for optical tissue pH determination.
• Reference tissue pH was measured using 750 um
  diameter micro-pH electrodes.
• Matching spectra and reference values modeled by
  multivariate calibration techniques. R2 and the
  standard error of cross-validation (SECV) used to
  assess model accuracy.

39
MODEL DEVELOPMENT

• Partial least-squares, factor analysis
• Create calibration with as many points as
  possible
• Cluster analysis
• Investigate (in)homogeneity of spectra

40
RESULTS
41
SPECTRA COLLECTION Lactate

• 82 raw absorbance spectra shown below (667-2400
  nm)
• Key features are water (970, 1450 and 2000 nm),
  cholesterol and its esters (1750 nm).

Thrombus/Red n22
Fatty/Yellow n41
Calcified/White n19
42
SPECTRA COLLECTION Tissue pH

• 48 raw absorbance spectra shown below (400-1100
  nm)
• Key features are hemoglobin (550 575 nm) and
  water (970 nm) absorption

Thrombus/Red n11
Fatty/Yellow n23
Calcified/White n14
43
Histogram of lactate reference measurements 3.2 ?
2.7 (mean ? SD) n82
44
Histogram of tissue pH reference
measurements 7.33 ? 0.21 (mean ? SD) n48
45
REFERENCE MEASUREMENTS

• No spurious correlations between measured
  variables
• Tissue temperature, experiment times within
  validated experiment parameters
• Pathology subjective

46
RESULTS Tissue Lactate

• 6-Factor model from 17 points
• Wavelength regions contributing to model
• 2030 2330 nm
• The R2 of the determination for optical lactate
  (LA) calibration 0.83.
• Estimated accuracy 1.4 micromoles LA/gram
  tissue.

47
Clustering solution for 82 spectra for the
optical determination of lactate. Cluster A 45
spectra, B 31 spectra, and C 6 spectra.
Cluster A contained the first 21 calibration
spectra collected.
48
RESULTS Tissue pH

• 3-Factor model from 17 points
• Wavelength regions contributing to model
• 1 400 615 nm
• 2 925 1890 nm
• 3 2044 2342 nm
• The R2 of the determination for optical tissue pH
  calibration 0.75.
• Estimated accuracy 0.09 pH units.

49
Clustering solution for optical determination of
tissue pH. Two clusters A contains 39 spectra,
B contains 9 spectra. The underlying pathology in
cluster B was identified as all thrombotic
points.
50
DISCUSSION

• Lactate model on portion of entire data set
• Further factor analysis showed spectra weakly
  associated with theoretical lactate peaks
• Number of factors in model too high for of
  samples used need more samples
• Tissue pH model on portion of entire data set
• Further factor analysis showed spectra associated
  with Hb and water peaks, evidence of pH-induced
  shift
• Number of factors acceptable for of samples used

51
CONCLUSIONS

• A small fiber optic prototype can make optical
  measurements in-vitro for the assessment of
  metabolically active plaque in a defined region
  of tissue (lt 1 mm3 volume).
• Hypothesis accepted
• in-vitro experimental factors can be assessed to
  their importance in affecting the optical
  calibration accuracy. The tissue temperature,
  experiment time, and gross pathology are
  identified a priori.
• Hypothesis accepted

52
CONCLUSIONS

• Mathematical models can be developed which relate
  the corresponding optical spectra to the
  individually measured metabolic parameters
  (tissue pH and lactate concentration) in the
  presence of inherent pathological variability.
• Hypothesis rejected for large n pending work
• Limited feasibility of models generated
• Pathological variability large
• Unmodeled tissue variability
• Lactate reference method precision
• Optical tissue volume gtgt real tissue pH
  heterogeneity
• Long-term spectrometer drift could not be ruled
  out

53
FUTURE WORK

• Considerable in-vitro work needs to continue
• Other clustering algorithms, pre-processing
  methods
• Reference lactate measurement precision
• Reduce unmodeled variability, better tissue model
• Larger data sets
• Spectrometer stability
• Rigorous acceptance criteria must be met before
  use in-vivo animals or humans

54
ACKNOWLEDGEMENTS
Texas Heart Institute, Center for Vulnerable
Plaque Research/UT Houston Dr. S.W. Casscells
Dr. Silvio Litovsky Dr. Morteza
Naghavi Department of Surgery, University of
Massachusetts Medical School Vascular Surgeons
Dr. P. Nelson, Dr. B. Cutler, Dr. A. Fox and Dr.
M. Rohrer Dr. Babs R. Soller Dr. Patrick
Idwasi This work was supported by US Army DREASM
Grant