Functional Partitioning of the Human Lumbar Multifidus: an Analysis of Muscle Architecture, Nerve and Fiber Type Distribution Using a Novel 3D in Situ Approach

Home , Multifidus muscle

Functional Partitioning of the Human Lumbar Multifidus: An Analysis of Muscle Architecture, Nerve and Fiber Type Distribution using a Novel 3D in Situ Approach

Alessandro Rosatelli

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Institute of Medical Science University of Toronto

Functional Partitioning of the Human Lumbar Multifidus: An Analysis of Muscle Architecture, Nerve and Fiber Type Distribution using a Novel 3D in situ Approach

Alessandro L Rosatelli

Doctor of Philosophy

Institute of Medical Science University of Toronto

2010

Abstract

Muscle architecture, innervation pattern and fiber type distribution of lumbar multifidus

(LMT) throughout its volume was quantified. Musculotendinous (n=10) and neural

components (n=3) were dissected and digitized from thirteen embalmed cadaveric

specimens. The data were imported into Autodesk® Maya® 2008 to generate 3D

neuromuscular models of each specimen. Architectural parameters (fiber bundle length,

FBL; fiber bundle angle, FBA; tendon length) were quantified from the models using

customized software. The medial branch of the posterior rami (L1-L5) was traced

through LMT to determine its distribution. Using immunohistochemistry, Type I/II

muscle fibers were identified in 29 muscle biopsies from one fresh frozen specimen. The

total area and number of each cell type was calculated using Visiopharm® (image analysis

software). Architectural and fiber type data were analyzed using ANOVA with Tukey’s post-hoc test (p ≤ 0.05).

ii From L1-L4, LMT had three architecturally distinct regions: superficial, intermediate and deep. At L5, intermediate LMT was absent. Mean FBL decreased significantly from superficial (5.8 ± 1.6cm) to deep regions (2.9 ± 1.1cm) as did volume

(superficial, 5.6 ± 2.3ml; deep, 0.7 ± 0.3ml). In contrast, mean FBA increased from superficial to deep. The medial branch of the posterior ramus (L1-L5) supplied the five bands of LMT. Each medial branch in turn divided to supply the deep, intermediate and superficial regions separately. The area occupied by Type I fibers was significantly less

(p< 0.01) in the deep (56%) compared with the superficial regions (75%).

Based on architecture and morphology, superficial LMT with the longest FBL and relatively small FBA is well designed for torque production and controlling the lumbar lordosis. Intermediate LMT with significantly longer FBL compared with the deep region and with its caudal to cranial line of action may help to control intersegmental stability. Furthermore, the absence of intermediate LMT at L5 and may contribute to the higher incidence of instability observed at the lumbosacral junction. Deep LMT with its short FBL, large FBA and proximity to the axis of spinal rotation may function to provide proprioceptive input to the CNS rather than a primary stabilizer of the lumbar spine.

iii Acknowledgments

If I have achieved anything thus far it is because I had the opportunity to work with some truly great people. First and foremost I would like to thank my mentor and research supervisor Dr. Anne Agur who has provided me support, advice and encouragement throughout my graduate career. She has challenged me, pushed me and made me strive far beyond what I perceived to be possible.

My sincerest thanks go to my advisory committee consisting of Dr. Bernie

Liebgott, Dr. Karan Singh, and Dr. Sharon Switzer-McIntyre for their expert advice, assistance and guidance. I would like to recognize the participation of both my internal and external examiners, Dr. Scott Thomas and Dr. Thomas Quinn. I also extend my gratitude to Dr. Mike Wiley and Dr. Ian Taylor for reviewing this thesis and providing much appreciated feedback.

I thank my fellow graduate students in the Division of Anatomy, Department of

Surgery: Soo Kim, Christopher Yuen, as well as Kajeandra and Mayo Ravichandiran who were instrumental in software development and implementation.

Thanks go to the anatomy technical staff of Bill Wood, Terry Irvine, and Jerry

Topham for their expertise in preparing the cadaveric specimens used in my graduate studies. Marianne Rogers at Mt. Sinai Hospital, spent many hours teaching me how to use the image analysis software used to examine muscle biopsy specimens, I extend my deepest gratitude. Your assistance was greatly valued.

Most importantly, I would like to thank my family for their unwavering love and support. Their constant encouragement lifted my spirits and lightened the journey, particularly when it was needed most. A special dedication goes to my parents from

iv whom I drew strength by emulating their perseverance, desire, and dedication. Lastly, to my wife Andrea, my best friend and most critical reviewer, thank you for believing in me.

Acknowledgement is made to the AO/ASIF Foundation, Switzerland and the

Department of Surgery, University of Toronto for financial support.

v Table of Contents

Page

Abstract ...... ii

Acknowledgments...... iv

Table of Contents...... vi

List of Figures ...... xi

List of Tables...... xiv

List of Abbreviations ...... xv

Chapter 1: Introduction...... 1

1.1 Contents of Thesis...... 5

Chapter 2: Literature Survey ...... 6

2.1 Muscle Architecture...... 6

2.1.1 Overview of Architectural Parameters...... 6

2.1.1.1 Fiber Bundle Length (FBL) ...... 9

2.1.1.2 Fiber Bundle Angle (FBA) ...... 11

2.1.1.3 Muscle Volume and Mass...... 11

2.1.1.4 Physiological Cross Sectional Area (PCSA) ...... 12

2.1.1.5 Fiber Type Distribution ...... 13

2.1.2 Functional Significance of Muscle Architecture ...... 14

2.2 Why Study Human Lumbar Multifidus Architecture? ...... 17

2.3 Previous Studies on the Morphology, Architecture, Innervation and Fiber Type distribution of LMT ...... 19

2.3.1 Morphology of LMT...... 19

2.3.2 Architecture of LMT...... 20

2.3.2.1 Qualitative/Descriptive Studies ...... 20

2.3.2.2 Quantitative Cadaveric Studies and their Results...... 24

2.3.2.2.1 Quantification of FBL...... 24

2.3.2.2.2 Quantification of FBA ...... 25

2.3.2.2.3 Quantification of Volume ...... 29

2.3.3 Innervation of LMT ...... 29

2.3.3.1 Motor Control of Lumbar Stability...... 30

2.3.4 Fiber Typing of LMT...... 31

2.3.4.1 Cadaveric Investigations...... 32

2.3.4.2 In Vivo Investigation ...... 33

2.3.4.3 Comparison of Cadaveric and In Vivo Measurements...... 34

2.4 Functions of LMT based of Morphological, Biomechanical, Electromyographic and Clinical Evidence...... 35

2.4.1 Morphological Evidence...... 35

2.4.2 Biomechanical Evidence...... 37

2.4.2.1 Role of LMT in Torque Production...... 37

2.4.2.2 Role of LMT in Spinal Stability ...... 39

2.4.2.2.1 Role of LMT in Maintaining the Lumbar Lordosis...... 39

2.4.2.2.2 Role of LMT in Controlling Shear Forces...... 40

2.4.2.2.3 Biomechanical Models for the Stability Role...... 40

2.4.2.2.4 Role in Providing Stiffness to the Spine...... 44

2.4.3 Electromyographic Evidence...... 45

2.4.3.1 LMT Activity Involved in the Maintenance of Posture...... 46

vii 2.4.3.2 LMT Activity in Active Lumbar Movements ...... 46

2.4.3.3 LMT Activity During Internal and External Perturbations of the Trunk...... 48

2.4.4 Clinical Evidence...... 49

2.4.5 Summary...... 50

Chapter 3 Hypothesis and Objectives...... 51

3.1 Hypotheses...... 51

3.2 Objectives ...... 51

3.3 Significance...... 52

Chapter 4 Methods ...... 53

4.1 Digitization, modeling and quantification of architectural parameters of LMT ...... 53

4.1.1 Specimens ...... 53

4.1.2 Serial dissection and digitization of muscle fiber bundles, tendons and spinal column ...... 53

4.1.3 Microscribe® 3G2 Digitizer ...... 57

4.1.4 3D reconstruction and modeling of LMT ...... 58

4.1.5 Quantification of architectural parameters of LMT...... 59

4.1.5.1 Fiber Bundle Length (FBL) ...... 59

4.1.5.2 Tendon Length...... 59

4.1.5.3 Fiber bundle angle (FBA) ...... 60

4.1.5.4 Volume...... 61

4.1.6 Statistical analysis of architectural parameters...... 61

4.2 Digitization and modeling of the intramuscular nerve distribution of LMT ..... 62

4.2.1 Specimens ...... 62

viii 4.2.2 Serial dissection and digitization of the medial branch of the posterior rami of L1 to L5...... 62

4.2.3 Reconstruction, modeling and analysis of intramuscular nerve distribution ...... 65

4.3 Fiber typing of LMT ...... 66

4.3.1 Specimen(s)...... 66

4.3.2 Sectioning and Immunohistochemistry...... 68

4.3.3 Morphometric analyses of Type I/II fibers ...... 69

4.3.4 Statistical analysis...... 70

Chapter 5 Results...... 72

5.1 Morphology and Architecture of Lumbar Multifidus...... 72

5.1.1 Superficial LMT...... 72

5.1.2 Intermediate LMT...... 73

5.1.3 Deep LMT...... 73

5.1.4 Architectural parameters...... 76

5.1.5 Tendon architecture ...... 79

5.2 Innervation of LMT ...... 80

5.2.1 3D Model ...... 80

5.2.2 Nerve distribution through LMT ...... 83

5.3 Mean characteristics of muscle fiber type for LMT: Pilot Study ...... 92

5.3.1 Fiber type distribution...... 92

5.3.2 Fiber type diameter ...... 96

Chapter 6 Discussion ...... 100

6.1 Introduction...... 100

ix 6.2 Morphology and Architecture...... 101

6.2.1 Morphology...... 102

6.2.2 Measurement of architectural parameters of LMT ...... 104

6.2.2.1 Fiber Bundle Length (FBL) ...... 105

6.2.2.2 Fiber Bundle Angle (FBA) ...... 107

6.2.2.3 Volume...... 110

6.2.2.4 Physiological Cross Sectional Area (PCSA) ...... 111

6.3 Innervation ...... 114

6.4 FiberType Characteristics ...... 115

6.4.1 Fiber Type Distribution...... 118

6.4.2 Fiber Size ...... 122

6.5 Functional considerations ...... 123

6.6 3D Reconstruction and Modelling: Pros and Cons...... 127

Chapter 7 Conclusions ...... 129

7.1 Functional paradigm ...... 130

Chapter 8 Future Direction ...... 132

Chapter 9 References ...... 134

x List of Figures

Page

Figure 2.1. Muscle architectural parameters...... 12

Figure 2.2. Length-force curve of two muscles with different PCSAs but equal FBL...... 16

Figure 2.3. Length-force curve of two muscles with different fiber lengths but equal FBA and PCSAs...... 17

Figure 2.4. Drawing of the cervical vertebrae showing how the cervical musculature stabilizes the cervical spine similar to guywires stabilizing the mast of a ship...... 20

Figure 2.5. Illustrations of the fascicles of lumbar LMT as seen in a posterior- anterior view...... 23

Figure 2.6. Posteroanterior view of the mean FBAs of the various fascicles of multifidus from the Ll to L5 spinous processes...... 26

Figure 2.7. Lateral view of the mean orientation of the fascicles of the multifidus from the Ll to L5 spinous processes...... 27

Figure 2.8. Posterior view of lumbo-sacral spine showing typical orientation of fascicle of LMT...... 37

Figure 4.1. Digitization of human LMT...... 54

Figure 4.2. Delineation of muscle fiber bundle of LMT (left, lateral view of lumbosacral spine)...... 55

Figure 4.3. Close up, lateral view of LMT originating from the L1 spinous Process showing a small segment of tendon (left, lateral view of spine). ... 56

Figure 4.4. Right, lateral view of digitized lumbar spine and sacrum as viewed in Autodesk® Maya®...... 57

Figure 4.5. The Immersion Company Microscopic 3-G2 Digitizer...... 57

Figure 4.6. Measurement of muscle fiber bundle length and tendon length of LMT (right, lateral view of lumbosacral spine)...... 59

Figure 4.7. Calculation of muscle fiber bundle angle (right, lateral view of lumbosacral spine)...... 60

xi Figure 4.8. Measurement of muscle volume...... 61

Figure 4.9. Flowchart outlining the process of serial dissection and digitization of the medial branch of the posterior ramus...... 64

Figure 4.10. Digitization of bony skeleton, 3D reconstruction...... 65

Figure 4.11. Lateral view of LMT showing deep (purple), intermediate (yellow) and superficial (red) regions...... 67

Figure 4.12. Typical microscopic view of the transversely sectioned LMT muscle...... 68

Figure 5.1. Digitization and three dimensional modeling of superficial, intermediate and deep regions of lumbar multifidus (LMT) of a cadaveric specimen, lateral views...... 74

Figure 5.2. Digitization and three dimensional modeling of superficial segments of lumbar multifidus (L1-L5), of a cadaveric specimen, lateral views...... 75

Figure 5.3. Histogram of fiber bundle length (FBL)...... 79

Figure 5.4. Views of the nerve supply to lumbar multifidus (LMT) by rotation of model...... 81

Figure 5.5. Views of the nerve supply to lumbar multifidus (LMT) at different magnifications...... 82

Figure 5.6. Views of the nerve supply to specific regions of lumbar multifidus (LMT)...... 82

Figure 5.7. Lateral view of lumbosacral spine showing medial branches (L1 to L5) which supply the five bands of LMT………………………………………86

Figure 5.8. Lateral view of the lumbar spine showing the extramuscular course of the L1 medial branch (solid blue line) traversing the intersection (shaded blue area) formed between the transverse process (tp) and superior articular process (sap) of L2...... 87

Figure 5.9. Dissection of lumbar multifidus showing extramuscular course of L1 medial branch, right lateral view...... 87

Figure 5.10. Close up lateral view of the L2 lumbar vertebra and L1 medial branch. .... 88

Figure 5.11. Dissection of lumbar multifidus, right lateral view showing main trunk of L1 medial branch (yellow dotted line) giving off a nerve branch to supply deep LMT (a: red dotted line)...... 88

xii Figure 5.12. Lateral view of lumbar spine showing L1 medial branch giving off articular branch (solid red line) to supply the superior zygapophyseal joint...... 89

Figure 5.13. Lateral view of the lumbar spine showing medial branch dividing into three branches...... 89

Figure 5.17. Lateral view of lumbosacral spine showing the L5 medial branch innervating deep and superficial regions of LMT...... 91

Figure 5.18. Mean area of Type I fibers for each region expressed as a proportion...... 93

Figure 5.19. Comparison of the mean area occupied by Type I fibers between the deep, intermediate and superficial regions of LMT...... 94

Figure 5.20. Mean areas of Type I fibers expressed as a proportion...... 95

Figure 5.21. The relationship between mean Type I cell diameters and spinal level...... 97

Figure 5.22. The relationship between mean Type I fiber diameters and spinal level and region...... 98

Figure 5.23. The relationship between mean Type II fiber diameters and region...... 99

Figure 6.1. Lateral view of lumbosacral spine showing superficial and intermediate regions of LMT attaching to the L1 spinous process...... 104

Figure 6.2. Bar graph showing comparison of FBL values as a function of segmental level...... 107

Figure 6.3. The net action of a given fascicle of LMT is dependant on the number and distribution of muscle fibers bundles...... 109

xiii List of Tables

Page

Table 2.1. Muscle Length and fiber bundle length of selected upper and lower limb muscles…………………………………………………………………… 10

Table 2.2. Architectural Data from Previous Studies………………………………… 28

Table 4.1. Spatial distribution of muscle biopsies taken from specimen……………… 67

Table 5.1. Summary of Mean FBL for LMT………………………………………….. 76

Table 5.2. Summary of Mean FBA for LMT………………………………………… 77

Table 5.3. Summary of Mean Volume for LMT……………………………………… 78

Table 5.4. Tendon length, FBL and muscle lengths of superficial and intermediate regions……………………………………………………………………… 80

Table 5.5. LMT Type I fiber proportions (mean ± SD)………………………………. 92

Table 5.6. LMT fiber type diameters (mean ± SD)……………………………………. 96

Table 6.1. Comparison of LMT muscle fiber angles of different studies including the current……………………………………………………………………… 109

Table 6.2. Comparison of PCSA of the current and previous studies………………... 113

xiv List of Abbreviations

2D Two dimensional m Mass 3D Three dimensional ML Muscle length ALP Alkaline Phosphatase mp Mammillary process BSA Bovine serum antibody MRI Magnetic resonance imaging CNS Central nervous system MVC Maximal voluntary contraction CSA Cross sectional area N Newton CT Computed tomography NA Not applicable ΔL Change in length of muscle fiber NZ Neutral zone δ l Change in length of sarcomere PCSA Physiological cross sectional area EMG Electromyography PSIS Posterior superior iliac spine F Force θ FBA FBA Fiber bundle angle S1 1st Sacral vertebra FBL Fiber bundle length Sa Sacrum GNP Gross national product SD Standard deviation HCI Hydrochloric acid sp Spinous process L Lumbar TL Tendon length LBP Low back pain tp Transverse process LMD Least mean diameter V Volume LMT Lumbar multifidus ZJ Zygapophyseal joint

xv Chapter 1 Introduction

Studies examining the microscopic properties of skeletal muscle fibers have provided great insights into their function (Lieber & Friden, 2000). However, relatively little work has been done to quantify the architectural characteristics of human skeletal muscle. The architecture of a muscle consists of its external configuration and dimensions, and the internal arrangement and morphology of the contractile and connective tissue elements. Even though two muscles may have the same external configuration they may differ significantly in function due to differences in the internal arrangement of their contractile and connective tissue elements (Lieber & Friden, 2000).

Why is it important to study muscle architecture? Put simply, it is an important determinant of function (Roy & Ishihara, 1997; Lieber & Friden, 2000). For example, physiological cross-sectional area (PCSA) is calculated from the architectural parameters of a muscle and is considered to be directly proportional to the maximum force or tension that can be generated by that muscle (Gans, 1982). In contrast, fiber length is proportional to fiber excursion and determines the range of lengths over which a muscle can generate active force (Zajac, 1989).

Visualization of muscle architecture and geometry has for the most part relied on data collected in 2D, either from cadaveric specimens (Friederich & Brand, 1990)or from imaging techniques, such as ultrasound (Maganaris et al., 1998; Chow et al., 2000; Martin et al., 2001). It is only recently due to advances in the area of computer modeling and digitization that the study of skeletal muscle architecture has taken a three dimensional perspective (Agur et al., 2003; Kim, Boynton et al., 2007).

1 2

One area which has received relatively little attention in regard to both the quantification and visualization of muscle architecture has been muscles of the back, despite the high incidence of disability due to low back pain (LBP) in the general population (Anderson, 1999). The lifetime prevalence of LBP has been estimated to range anywhere from 70% to 85% (Anderson, 1999). In any one year, the incidence of back pain is reported to affect approximately 5% of the population. In fact, it is estimated that in Europe and the United States at least 1% to 2% of the gross national product

(GNP) is allocated towards the management of this condition (Norlund & Waddell,

2000). In the United States this equates to approximately 1-2 billion dollars. Studying the detailed architecture of the back muscles may lead to better biomechanical models which would ultimately improve our understanding of LBP.

Cholewicki and McGill (1996) and Crisco and Panjabi (1991) have demonstrated the vital role of the deep, local muscles in controlling spinal stiffness. One such muscle thought to be pivotal in this regard is the lumbar multifidus (LMT). LMT is “the most medial of the back muscles and is the largest muscle that spans the lumbosacral junction.

Due to its prominence in this region, it is a preferred target for diagnostic paraspinal electromyography (EMG), and has been the subject of histochemical studies of patients with lumbar disorders” (Macintosh et al., 1986). In addition, alteration in neuromuscular recruitment of this muscle following acute LBP has been postulated to predispose the lumbar spine to further injury and instability (Hides et al., 1996; Hides et al., 2001).

However, due to the lack of architectural data available on LMT, the precise nature, distribution and magnitude of forces exerted on the lumbar spine by this muscle is

3 unavailable. Hence the functional role(s) of LMT within the lumbar spine remains largely unknown.

Fortunately advances in the area of computer graphics and modeling now make it possible to collect large quantities of morphological data to reconstruct the 3D architecture of skeletal muscle in situ (Ng-Thow-Hing, 2001; Agur et al., 2003; Kim,

Boynton et al., 2007). Using these techniques of data acquisition and modeling, the current study examines the morphology and architecture of LMT throughout its entire volume.The results obtained may help explain, among other things, how the fibers of

LMT work collectively to provide multidirectional movement to the lumbar spine while also providing stability.

Making matters even more complex, is the notion that certain muscles (e.g. supraspinatus) can be divided into architecturally distinct regions defined by fiber bundle orientation i.e. fiber bundle length, pennation angle (or fiber bundle angle) and tendinous attachments (Kim, Boynton et al., 2007) The unique arrangement of fiber bundles suggests that some muscles like supraspinatus may be composed of “neuromuscular compartments” which have specific, task-oriented roles. Neuromuscular compartments are defined as architecturally distinct regions within the muscle which are independently innervated by an individual nerve branch. Each compartment contains motor unit territories with a unique array of physiological attributes (English & Letbetter, 1982a). In other words, the intensity and timing of motor unit activation of each compartment can be independently controlled and can vary between regions. English & Letbetter (1982b) showed that the distribution of fiber types (Type I and Type II) within a muscle can vary

4 between compartments. Thus, the endurance and force producing abilities of individual muscle compartments may differ.

Evidence suggests that LMT provides stability to the lumbar spine but also mobilizes and produces movement. This functional duality suggests that LMT like supraspinatus is made up of neuromuscular compartments, with each compartment subserving a particular function. To substantiate the presence of neuromuscular compartments within LMT requires a thorough investigation of the innervation and fiber type distribution pattern throughout the volume of the muscle. To date however neither has not been investigated fully. Hence, there is insufficient morphological and histochemical evidence to support the hypothesis that LMT consists of architecturally distinct regions which are independently innervated by a single motor branch. Nor do we know if the fiber type distribution through the volume of LMT differs in any way from superficial to deep or from superior to inferior.

In this study, the muscle architecture, innervation and fiber type distribution of the human LMT has been documented throughout its entire volume. Ultimately, these data may lead to the development of more robust, biomechanical models to help elucidate the relationship between structure and function as it applies to the human LMT. With a better understanding of the 3D architecture of the LMT, including its nerve and fiber type composition, new insights into the way in which this muscle helps to protect and stabilize the spine during activities of daily living can be developed

1.1 Contents of Thesis

The present thesis consists of eight chapters presented in the following sequence.

Chapter 1 provides introductory background information on the importance of skeletal muscle architecture, nerve supply and fiber type distribution to muscle function.

Chapter 2 is the literature survey which is intended to provide background information on the structure and function of skeletal muscle, muscle modelling and imaging modalities including ultrasound and magnetic resonance imaging used to study human muscle in vivo. The existing literature of the architecture and functions of the cadaveric and in vivo human LMT is reviewed in detail.

Chapter 3 includes the hypotheses, objectives and significance of the study. The anatomy and terminology used to describe LMT in the present thesis are also defined.

Chapter 4 outlines the methods that are used to address the hypotheses and objectives of this thesis.

Chapter 5 is a summary of the results. This section is divided into three parts.

The first section reports both morphological and architectural data on LMT including visualization of the complex fiber bundle arrangement of this muscle in situ using 3D computer modelling. The next section documents the detailed course of the medial branch of the posterior ramus through the volume of LMT. The final section reports on the distribution of Type I and II muscle fibers throughout the volume of the muscle.

Chapter 6 is a discussion of the results and innovations of this thesis.

Chapters 7 and 8 consist of the conclusions and future directions of this work.

Chapter 2 Literature Survey

This literature survey provides background information for this comprehensive study of the muscle architecture, nerve and fiber type distribution of LMT. Section 2.1 provides the reader with a general overview of skeletal muscle architecture, including discussion of architectural parameters important in defining and understanding muscle function. Section 2.2 discusses briefly the importance of studying the morphology and architecture of lumbar multifidus. Section 2.3 provides an overview of the morphology, nerve supply, fiber type distribution and action(s) of LMT. A summary of previous descriptive and quantitative studies of LMT architecture, including previous data on FBL,

FBA and muscle volume (V) are also included in this chapter. Lastly, in Section 2.4, studies which have investigated the function(s) of lumbar LMT are discussed.

2.1 Muscle Architecture

2.1.1 Overview of Architectural Parameters

The arrangement of muscle fibers (i.e. muscle architecture), is a primary determinant of muscle function (Lieber & Friden, 2000). Hence, understanding how muscle structure influences muscle function is of significant scientific and clinical importance. This structure-function relationship is essential for the following reasons:

“Clarifies the physiological basis of force production, movement” (Lieber &

Friden, 2000) and stability (Panjabi et al., 1989; Panjabi, 1992b, 1992a; Wilke

et al., 1995; Cholewicki & McGill, 1996).

6 7

Provides data for the proper placement of electromyographic surface

electrodes with respect to the muscle fiber direction which is critical for

obtaining valid measures of muscle activity (De Foa et al., 1989),

Helps infer the “mechanical basis of muscle injury during normal movement”

(Lieber & Friden, 2002), and

Assists in the “interpretation of histological makeup of specimens obtained

from muscle biopsies” (Roy et al., 1991).

Despite the importance of muscle architecture, little attention has been directed towards quantifying many of its associated parameters. These include measures of: fiber bundle length (FBL), muscle length (ML), fiber bundle angle (FBA) (also used interchangeably with pennation angle), muscle volume, density, physiological cross sectional area (PCSA) and fiber type distribution. Most anatomical text books typically depict the macroscopic structure of skeletal muscle as collections of muscle fiber bundles

(i.e. fascicles) projecting from a point of “origin”, to a point of “insertion”. This oversimplification of skeletal muscle architecture does not adequately capture the in situ, complex, three dimensional nature of this highly organized tissue. The architecture of a given muscle has been shown to be relatively consistent between members of the same species (Lieber & Friden, 2000). This being said, the function of a particular muscle may be altered by pathology, injury or disease processes, particularly if the macroscopic arrangement of its muscle fibers are affected. Therefore, it is essential to clearly establish the normal structure and architectural parameters of skeletal muscle tissue in order compare and contrast the same parameters examined in abnormal skeletal muscle tissue.

It has been argued that the architectural properties of FBL, muscle length, and

FBA are the “most structurally significant parameters, whereas FBL, muscle length, and fiber type distribution are the most functionally determining” (Burkholder et al., 1994).

In addition, architectural differences between muscles have been shown to result in significant effects on tension and contractile speed among muscles despite having the similar fiber type (Gans, 1982; Sacks & Roy, 1982; Roy et al., 1984). In this way, the contractile properties of muscle can be modulated by changing the muscle’s architectural properties. For example, Mohagheghi et al (2007) used ultrasonography to assess, in vivo, the gastrocnemius muscle architecture in the paretic and non-paretic legs of eight children with cerebral palsy. They found that fiber bundle/fascicle length and muscle thickness were reduced by up to 18% and 20% in the paretic compared to the non-paretic legs respectively. The authors concluded that paresis in hemiplegic cerebral palsy may affect the geometry of skeletal muscle which in turn will alter its function.

Ward et al (2006) examined the architectural properties of the rotator cuff muscles in 10 cadaveric specimens and concluded these muscles are important in maintaining glenohumeral stability both at rest and in end range positions. These authors also suggested small changes in rotator cuff muscle length that can occur as a result of surgery, may result in relatively large changes in shoulder function.

Kim, Boynton et al (Kim, Boynton et al., 2007) studied the muscle architecture of the supraspinatus muscle in 10 cadaveric specimens. The authors found that the muscle belly of supraspinatus could be divided into anterior and posterior regions. Each region, in turn, could be further subdivided into superficial, middle, and deep parts. The significantly larger muscle volumes associated with the anterior region when compared

9 with the posterior region were suggested to influence the amount of force generated by each region, with the anterior region generating the majority of the muscle’s tensile force.

The significant differences in fiber bundle angle found among different parts of the anterior region of the muscle was postulated to result in a heterogeneous distribution of forces and hence influence the higher occurrence of deep, articular tears of the anterior tendon with rotator cuff pathology.

Based on the aforementioned evidence, muscle architectural parameters such as

FBL, FBA, ML, PCSA, etc…directly influence muscle function and pathology.

However, before proceeding any further, it is important to define these variables.

2.1.1.1 Fiber Bundle Length (FBL)

FBL is the length of a muscle fiber bundle from its most proximal end to its most distal end (Figure 2.1). The length of a muscle fiber is determined by the number of sarcomeres arranged end to end. Within one muscle fiber bundle the sarcomere length remains quite consistent (Wickiewicz et al., 1983). When a muscle contracts, each sarcomere shortens proportionately (Alberts et al., 1989), resulting in shortening of the muscle fiber by approximately one third of its length (Enoka, 1988). This relationship can be expressed as follows:

ΔL is the change in length of the muscle fiber Δ=L nl()δ n is the number of sarcomeres in series

δ l is the change in length of a sarcomere

Therefore, a muscle with more sarcomeres in series undergoes the greatest absolute change in length thereby resulting in greater muscle excursion or degree of shortening.

Muscle length (ML) in comparison to FBL is defined as “the distance from the origin of the most superior muscle fibers to the insertion of the most inferior fibers”

(Lieber, 2002). Under most circumstances, ML is greater than FBL (Table 2.1); however, as muscle fibers become oriented more parallel to the force generating axis of the muscle, these two values approach one another (Table 2.1). For example the muscle fibers of the brachioradialis muscle are oriented almost parallel to the force generating axis of the muscle and its ML (175 mm ± 8.3mm) is approximately equal to its FBL

121mm ± 8.3mm.

Muscle Muscle Length (mm) Fiber Bundle Length Brachioradialis (BR) 175 ± 8.3 121 ± 8.3 Biceps Femoris (BF) 271 ± 11 139 ± 3.5 Flexor Pollicis Longus (FPL) 168 ± 10.0 45.1 ± 2.1 Medial Gastrocnemius (MG) 248 ± 9.9 35.3 ± 2.0 Popliteus (POP) 108 ± 7.0 29.0 ± 7.0 Pronator Quadratus (PQ) 39.3 ± 2.3 23.3 ± 2.0 Pronator Teres (PT) 130 ± 4.7 36.4 ± 1.3 Rectus Femoris (RF) 316 ± 5.7 66.0 ± 1.5 Sartorius (SAR) 503 ± 27 455 ± 19 Semimembranosus (SM) 262 ± 1.5 62.7 ± 4.7 Soleus (SOL) 310 ± 1.5 19.5 ± 0.5 Vastus Lateralis (VL) 324 ± 14 65.7 ± 0.88 Vastus Medialis (VM) 355 ± 15 70.3 ± 3.3 Table 2.1. Muscle Length and fiber bundle length of selected upper and lower limb muscles (Lieber, 2002).

2.1.1.2 Fiber Bundle Angle (FBA)

In general for skeletal muscle, FBA is defined as the angle subtended between a muscle fiber and its force-generating axis. It is usually measured by determining the average angle of fibers on the superficial surface of the muscle. When the fibers run essentially parallel to the line of force, the muscle is said to be longitudinal. If the fibers are oriented at a single angle relative to the line of force then they are said to be unipennate. While, muscles that are made up of fibers that are oriented at several angles relative to the axis of force generation are termed multipennate (Figure 2.1). Generally, the greater the FBA, the smaller the amount of effective force transmitted to the tendon.

This relationship is represented using the following equation:

Fnet= F(muscle fiber bundle) cos θ

F represents force (Enoka, 1988) θ represents the FBA

As the fiber angle θincreases, the force transmitted to the attachment site will consequently decrease. Hence, using the cosine law, as θ approaches 90 degrees, the net force generated by the muscle fiber approaches zero.

2.1.1.3 Muscle Volume and Mass

Volume and mass can also be measured. Volume is usually measured using water displacement techniques and mass is measured as the wet weight of the muscle

(Friederich & Brand, 1990). If the volume and mass of a muscle are known, the density can be calculated using the following formula:

Figure 2.1: Muscle architectural parameters. (A) Longitudinal arrangement of muscle fiber bundles running parallel to the muscles’ force-generating axis. Note: Since muscle fibers run parallel to the axis with which force is generated (i.e. tendon), then the fiber bundle angle (FBA) is zero; (B) Unipennate arrangement of muscle fiber bundles. Fibers run at a fixed angle relative to the muscles’ force-generating axis. (C) Multipennate architecture in which muscle fibers run at several angles relative to the muscles’ force- generating axis. ML, muscle length; FBL, fiber bundle length; FBA, fiber bundle angle. (Adapted from Lieber and Friden, 2000, Figure 1 using Figures 5.23B, 5.23C, and 6.32 from Grant’s Atlas of Anatomy, 2005).

2.1.1.4 Physiological Cross Sectional Area (PCSA)

Perhaps one of the most important architectural parameters to quantify is PCSA.

It is usually derived from the more common architectural parameters discussed previously. This measure is important as it is directly proportional to the amount of force a muscle can generate. The PCSA of a muscle is represented by the formula:

Muscle Mass (g) * cosine 2 θ = fiber bundle angle PCSA (cm ) = ρ = muscle density ρ (g/cm3)* Fiber length (cm)

Estimations of muscle force are sometimes based on muscle volume, without taking into consideration the arrangement of muscle fibers (i.e. muscle architecture).

Using cross sectional areas of muscles obtained from Computed tomography (CT) or

Magnetic resonance imaging (MRI) images has typically led to erroneous estimations of force production and false extrapolations of muscle function (McGill et al., 1988).

Transverse sections taken of a muscle with CT or MRI do not capture the large number of muscle fibers that are potentially present in pennated muscles. As a result, muscle forces are commonly underestimated. Hence, areas obtained from CT or MRI must take into consideration the architecture of a muscle (McGill et al., 1993).

2.1.1.5 Fiber Type Distribution

The fiber type composition of a skeletal muscle is an important parameter to consider when determining the function of a muscle. There are two main fiber types:

Type I (slow twitch) and Type II (fast twitch). Although muscles typically have a mixture of both these fiber types, there is usually a predominance of one that influences the contractile properties of the entire muscle (Lieber, 2002). Type I fibers are found to predominate in postural muscles that are able to maintain repetitive contractions for long periods of time before fatigue occurs. Type II fibers fatigue more easily, but can contract about three times as fast as Type I fibers and are suited for generating high forces for short periods of time. Each fiber type has a different protein composition which can be altered by hormonal and neuronal factors as well as specific exercise training. These changes can result in subtle changes in the contractility of a muscle (Lieber, 2002).

2.1.2 Functional Significance of Muscle Architecture

Examination of mammalian muscles tissue reveals a complex internal architecture. Lieber and Blevins (1989) pointed out, “skeletal muscle contractile properties are a function of both the intrinsic muscle fiber properties and the fiber arrangement within the muscle (i.e., the architectural design)”. Hence, the determination of muscle function is more intricate than just considering ‘origin and insertion’, or assessing activity profiles with EMG. As Liber & Friden (2000) point out, muscle architecture has a significant impact on the functional properties of a muscle. In fact, architectural differences between muscles are excellent predictors of force generation

(Lieber & Friden, 2000).

The functional effect of muscle architecture can be simply stated as: “muscle force is proportional to the physiological cross sectional area PCSA, and muscle velocity and excursion are proportional to the fiber length” (van Eijden et al., 1995). Neither fiber length nor PCSA can easily be measured based on gross muscle inspection. Instead, detailed dissections of cadaveric muscles are required for architectural determination

(Lieber, 2002). Upon determining architectural properties, it is possible to understand how much force the muscle generates and how fast it contracts.

Take for example the human pterygoid muscles which have a complex architectural design. The lateral pterygoid is fan shaped with relatively long muscle fibers, while its counterpart, the medial pterygoid, is multipennated with short muscle fibers (Williams,

1995). Furthermore, the lateral head is composed of two separate heads or regions, a superior head and an inferior head. The superior head is thought to be activated during jaw closing, while the inferior head is activated during opening (Juniper, 1981; Wood et

15 al., 1986). The medial pterygoid is a primary elevator of the mandible (Friedman, 1988).

In order to determine the magnitude and degree of excursion associated with the lateral and medial pterygoid muscles architectural parameters such as FBL and PCSA must first be quantified experimentally. Using the results obtained from eight cadavers, van Eijden et al, 1995 showed that these parameters vary between the pterygoids. The lateral pterygoid is characterized by relatively long fibers and a small PCSA, whereas the medial pterygoid has relatively short fibers and a large PCSA. The mechanical consequence is that the lateral pterygoid can produce displacements and velocities that are 1.7 times larger than the medial pterygoid, whereas the medial pterygoid can produce forces that are about 1.6 times greater than the lateral pterygoid. Similar results have been found in other human skeletal muscles in both the upper (Lieber et al., 1990) and lower extremities

(Wickiewicz et al., 1983); (Friederich & Brand, 1990).

Two specific architectural examples and their impact on the length-tension and force-velocity relationships are illustrated below.

Assume that two muscles had identical FBL and FBA, but one muscle had twice the

PSCA. What effect would this have on the functional properties of the muscle? Figure

2.2 below demonstrates the only functional effect that would occur is an increase in maximum tension so that the length-tension curve would be the same basic shape but simply amplified upward in the case of the stronger muscle (Lieber, 2002).

Figure 2.2. Length-force curve of two muscles with different PCSAs but equal FBL. The muscle with the larger PCSA produces the greatest force output. (Adapted from Lieber and Friden, 2000, Figure 8).

On the other hand, if two muscles with identical PCSAs and fiber bundle angles had different fiber lengths the effect would be an increase in the muscle velocity (i.e. increase muscle excursion). The peak absolute force of the length-tension curves would be identical, but the absolute muscle active range would be different (Figure 2.3).

Figure 2.3. Length-force curve of two muscles with different fiber lengths but equal FBA and PCSAs. The effect is to increase the range through which a muscle can generate a contraction but with retention of the same peak or maximal force. (Adapted from Lieber and Friden, 2000, Figure 9).

2.2 Why Study Human Lumbar Multifidus Architecture?

The isolated osseoligamentous lumbar portion of the spine is inherently unstable and has been shown to buckle under compressive loads of only 90N. In marked contrast, competitive weightlifters have safely and routinely exceeded 20,000 N of compressive force without their spines showing signs of mechanical failure (McGill, 2007). The human spine can obviously tolerate a great deal of stress, but the question remains how?

Panjabi (1992b, 1992a) proposed a model of lumbar spinal stability which seemed to

18 provide a plausible explanation to this apparent paradox as well as an impetus for more research. In his model, he proposed that spinal stabilization is a function of the interaction of three systems: the osseoligamentous system, the muscular system and the neural control system. Unlike earlier models of the spine which focused only on the role of passive structures (e.g. ligaments and joints) in stabilizing the spine, Panjabi’s model highlighted the important role of muscles, especially the deep (local) muscles (e.g. lumbar multifidus) in controlling spinal stiffness.

Human muscle architectural data, particularly those pertaining to the lumbar back muscles, are incomplete and based on relatively small sample sizes (McGill et al, 1986).

In addition, the sites of data collection have not been identified precisely within the muscle (Yamaguchi et al., 1990). As a consequence, modeling studies which included skeletal muscle did so in a “nominal or abbreviated way, such as including only some of the back muscles or approximating the action of several muscles into a single force- equivalent” (Bogduk et al 1992). Although the vital role of the local muscles in spinal stabilization has been recognized (Crisco & Panjabi, 1991; Cholewicki & McGill, 1996),

“many anatomic features accepted in the modeling literature were found to be highly inaccurate…” (McGill & Norman, 1986). The greatest oversimplification appears to have been made in the representation of the trunk extensor muscles (e.g. lumbar multifidus and erector spinae). Although Hides et al (1996; 2001) found that the LMT is a key muscle which provides stability to the lumbar spine, 3D morphology and architecture of the muscle are still not well represented in the literature. By studying the structure and architecture of LMT, that is, the size, arrangement, and distribution of its muscle fibers,

19 we may better understand the extent to which this muscle helps control stability, generate movement, and/or provide proprioceptive feedback.

2.3 Previous Studies on the Morphology, Architecture, Innervation and FiberType distribution of LMT

2.3.1 Morphology of LMT

Lumbar LMT is the most medial of the back muscles and is the largest muscle that spans the lumbosacral junction (Macintosh et al., 1986). The LMT is described in anatomical textbooks (e.g. Williams, 1995)as being made up of a “number of fleshy and tendinous fasciculi1 which fill up the groove on either side of the lumbar spinous processes of the vertebrae”. These fasciculi extend from the posterior surface of the sacrum, as far caudally as the “fourth sacral foramen, from the aponeurosis of origin of the sacrospinalis from the medial surface of the posterior superior iliac spine, and from the posterior sacro-iliac ligaments” (Williams, 1995). Muscle fibers of LMT are described as having an oblique orientation, traveling superior-medial direction to insert onto the spinous process of one of the vertebrae above. These fasciculi vary in length: the most superficial and longest pass from either the sacrum or “one of the lumbar vertebrae to the third or fourth above; those medial to or next in order run from one vertebra to the second or third above; while the deepest connect two adjacent vertebrae”

(Williams, 1995).

The LMT is supplied by the medial branches of the posterior rami of L1 to L5

(Macintosh et al., 1986). Each medial branch crosses the vertebral lamina, deep to the muscle, embedded in a layer of fat that separates LMT from bone. Each medial branch

1 Fascicle = a collection of muscle fiber bundles.

20 further divides into several branches to supply separate fascicles of LMT. Each nerve innervates only the fascicles that arise from the spinous process or lamina of the vertebrae with the same segmental number as the nerve (Macintosh et al., 1986).

2.3.2 Architecture of LMT

2.3.2.1 Qualitative/Descriptive Studies

Leonardo da Vinci (1452-1519) was the first “anatomist/biomechanist” to recognize the importance of understanding the relationship between muscle structure and function. He was the first to hypothesize that stability of the cervical spine was imparted, at least to some degree, by the unique architectural design of the cervical musculature

(Figure 2.4). Similar to the way guywires help to stabilize the mast of a ship, the cervical musculature helps to prevent the cervical spine from buckling beneath the weight of the head.

Figure 2.4. Drawing of the cervical vertebrae showing how the cervical musculature stabilizes the cervical spine similar to guywires stabilizing the mast of a ship. This is a faithful photographic reproduction of an original two-dimensional work of art. The work of art itself is public domain.

Following da Vinci’s line of reasoning, Bergmark (1989), suggested that the trunk muscles can be divided in two subsystems: the global-mobilizing-system and the local- stabilizing-system. Muscles belonging to the local muscle system are thought to be deep, have a small moment arm, span only a few vertebral levels, and are positioned close to the spines’ central axis. On the other hand, muscles belonging to the global system are thought to be more superficial, have much larger moment arms which are capable of producing larger forces, and are further away from the spinal axis of rotation

(Richardson, 1999). At present the LMT muscle is commonly assigned to the local system whereas the longissimus and iliocostalis muscles, which form the erector spinae, are assigned to the global system (Richardson, 1999).

These and other earlyaccounts of muscle architecture are purely descriptive in nature. In addition, while anatomical text books are invaluable as educational tools to assist with muscle localization, they are limited to the use of two dimensional images or photographs to represent complex three dimensional skeletal muscle structures or architecture. In the case of LMT, many anatomy texts books report this muscle as being part of the transversospinalis group of deep back muscles (Agur & Grant, 2005). They describe its structure by showing that its muscle fiber bundles as passing obliquely from the sacrum, mammillary processes of the lumbar vertebrae, and aponeurosis of the erector spinae, to insert into the spinous processes approximately three segments higher (Agur &

Grant, 2005).

These descriptions of LMT, however, do not provide the necessary anatomical data with which to accurately and consistently perform electromyography or tissue biopsy analysis of patients with lumbar disorders. As a consequence of this, Macintosh et al

(1986) performed an investigation to determine the morphology of LMT in order to provide an accurate anatomical foundation for future electromyographic and pathological studies of this muscle. These researchers studied 12 cadavers by gross dissection identifying small collections of muscle fiber bundles at either the superior or inferior ends and then stripping these away carefully from the remaining muscle mass while taking note of their specific attachment patterns.

These researchers found that LMT consisted of five separate bands, with each band consisting of a series of fascicles (collections of muscle fiber bundles) which stem from the spinous processes and laminae of each lumbar vertebra. Macintosh et al (1986) reported that in each band, the deepest and shortest fiber bundles arise from the vertebral lamina, while the other fiber bundles arise from the spinous process. The fibers originating from the L1 to L4 vertebral lamina insert onto the mammillary processes of the vertebra two levels caudally (Figure 2.5A), while the L5 fibers insert onto the dorsal surface of the sacrum just above the first sacral foramen. The fibers which originate from the spinous processes are longer than the laminar fibers. Fiber bundles from a given spinous process insert onto mammillary process (when present) three, four and five levels inferiorly. The longest fibers from L1 insert onto the posterior superior iliac spine

(Figure 2.5B). The fibers from L2 insert onto the posterior superior iliac spine and an area of the iliac crest just inferior to the posterior superior iliac spine (Figure 2.5C).

Similarly, the longer fibers from L3 insert onto an area from the posterior superior iliac spine to the lateral edge of the third sacral vertebra (Figure 2.5 D). The fibers from L4 insert onto the sacrum in an area medial to the L3 area of insertion (Figure 2.5E), while the fibers from L5 insert into an area medial to the dorsal sacral foramina (Figure 2.5F).

The schematic illustrations and descriptive accounts of LMT provided by

Macintosh et al (1986) while important to our overall understanding of LMT, unfortunately, do not provide a three dimensional perspective of the muscle as it appears in situ. Furthermore, data important to elucidating muscle architecture such as FBL,

FBA, muscle volume, PCSA and so forth cannot be extracted or determined by means of these linear models. Indeed, few studies have been performed which have attempted to fill this void.

Figure 2.5. Illustrations of the fascicles of lumbar LMT as seen in a posterior- anterior view. A illustrates laminar fibers from L1 to L5. B-F illustrates the longer, more superficial fibers attaching to the L1-L5 spinous processes. (from Macintosh et al, 1986, Figure 2).

2.3.2.2 Quantitative Cadaveric Studies and their Results

LMT is structurally and architecturally complex. Despite its complexity, LMT and the other back muscles are commonly incorporated into biomechanical models in an

“abridged manner, reducing their actions more or less to a single force equivalent”

(Hansen et al., 2006). A great deal of architectural data is lost in the process; data which are inherently important in determining the nature, distribution and types of forces acting through the lumbar spine. The lost architectural data include measurements of FBA,

FBL and muscle volume. Hence, the ability to extrapolate the actions of the back muscles from these models is dependent on the accuracy of the architectural data used as input parameters (Hansen et al., 2006). The sections which follow summarize the architectural data currently available for LMT.

2.3.2.2.1 Quantification of FBL

In their paper entitled “A Universal Model of the Lumbar Back Muscles in the

Upright Position” Bogduk et al (1992) constructed a model of the lumbar back muscles incorporating 49 fascicles of the lumbar erector spinae and LMT. These authors were primarily concerned with developing a model of the back muscles incorporating every fascicle in order to represent the actions of each of these muscles and all their fascicles on the lumbar spine.

In theory, the maximum force that can be exerted by a muscle is proportional to its

PCSA. Thus, in the case of the back muscles, in order to determine the PSCA of each component fascicle of the erector spinae and LMT, one needs to have measured both volume and FBL (recall: PCSA = volume/length). Bogduk et al (1992) measured these

25 architectural parameters during the course of previous morphological studies of the muscles (Macintosh et al., 1986; Macintosh & Bogduk, 1991). The authors report that as they resected away intact fascicles belonging to LMT they would record the length of the muscle belly (to the nearest 5mm), and its volume (to the nearest 1 ml) by inserting it into a volumetric cylinder containing water and recording the volume of water displaced.

Accordingly, physiologic cross-sectional areas were calculated by dividing the volume of each fascicle by its length (Bogduk et al., 1992).

Values for FBL (Table 2.2) were reported by Bogduk et al (1992), but these results do not represent the true length of each fiber bundle or fascicle. Rather, they are the length of the fascicle as projected in lateral radiographs of the spine (Bogduk et al.,

1992).

No other evidence on FBL for LMT was found in the literature.

2.3.2.2.2 Quantification of FBA

Macintosh and Bogduk (1986) studied the orientation of the fascicles of the LMT in five cadaveric specimens and summarized the descriptive data by plotting the orientation of a total of eleven fascicles with respect to their vertebrae of origin. These authors then measured the angle of each fascicle with respect to a standard reference line through each vertebra in both anteroposterior and lateral radiographs of the lumbar spine.

In the anteroposterior view, the reference line was a vertical line through the lumbar spinous processes on each outline of the lumbar spinous processes on each radiograph tracing (Figures 2.6). On the lateral radiographs the reference line was drawn

26 perpendicular to the posterior surface of the vertebral body through the superior-inferior midpoint of the vertebral pedicle (Figure 2.7).

Figure 2.6. Posteroanterior view of the mean FBAs of the various fascicles of multifidus from the Ll to L5 spinous processes. Each figure is labeled with the angle of every fascicle at each segmental level (with respect to the sagittal plane). Four fascicles are shown at Ll, with fewer fascicles represented at lower levels (from Macintosh and Bogduk, 1986, Figure 3).

Figure 2.7. Lateral view of the mean orientation of the fascicles of the multifidus from the Ll to L5 spinous processes. The dotted line represents the reference line through the pedicle against which the angle of orientation was measured. The table summarizes the orientations of the shortest to the longest fascicles at each segmental level. (From Macintosh and Bogduk, 1986, Figure 4).

In two later studies by De Foa et al (1989) and Biedermann et al (1991) the fiber direction of longissimus, iliocostalis and multifidus were studied to establish reference data in order to more reliably place surface electrodes for the purpose of electromyography of muscle activity. Using photographs taken from the posterior view of the back of both male and female cadaveric specimens they reported the “muscle fiber direction”, that is, fiber bundle angle for the dorsal back muscles relative to an anatomical reference line and the spine. For LMT, the FBAs reported for male and female cadaveric specimens were 15.1° (range: 13.5°-18.0°) and 23.5° (17.5°-28.5°) respectively. The FBA results from previous studies are summarized in table 2.2 below.

Ant-Post Lateral Angle Muscle Fiber Muscle Fiber Fascicle Length Angle (°) (°) Angulation (°) Angulation (°) Level (cm) Macintosh Macintosh et al, De Foa et al, Biedermann et Bogduk et al, 1992 et al, 1986 1986 1989 al, 1991

L1-L4 14.8 ± 0.8 86.8 ± 1.5 11.1 L1-L5 15.0 ± 0.7 85.4 ± 0.6 14.6 L1-S1 12.6 ± 0.6 90.0 ± 0.7 17.7 L1- 101.2 ± 1.1 16.6 ± 0.9 19.0 Sacrum L2-L5 18.8 ± 1.1 85.8 ± 1.6 9.8 L2-S1 18.0 ± 1.0 86.2 ± 0.4 Male Female 12.4 L2- 103.0 ± 0.7 20.0 ± 1.6 15.1± 1.43 23.5 ± 4.5 15.4 Sacrum (13.5-18.0) (17.5-28.5) L3-S1 23.2 ± 1.1 88.4 ± 2.3 8.0 L3- 102.0 ± 1.4 19.6 ± 0.9 11.9 Sacrum L4- 93.6 ± 2.3 15.6 ± 0.9 7.3 Sacrum L5- 93.8 ± 0.8 5.4 ± 1.5 4.1 Sacrum Table 2.2. Architectural Data from Previous Studies

2.3.2.2.3 Quantification of Volume

Bogduk et al (1992) reported having calculating muscle volume for LMT in a previous study using water displacement (Macintosh & Bogduk, 1986) however the authors have not reported the original data.

2.3.3 Innervation of LMT

Whatever structure may be causing pain, it must invariably involve a neuroanatomical physiological pathway. The lumbar spine is certainly no exception to this rule. Back pain is the second most common cause of sick leave in the United States, next only to the common cold (Guo et al., 1999). In 1990, it was estimated that 50-100 billion dollars were allocated to the treatment of this medical enigma (Guo et al., 1999).

Knowing the innervation of the lumbar spine is particularly important to surgeons performing minimal invasive surgery given their goal of preserving as much of the nerve supply as possible and to improving surgical outcomes.

Historically, the study of the nerve supply to the lumbar spine focused on descriptions of the sinuvertebral nerve, the lumbar dorsal rami and nerve supply of the lumbar zygapophyseal joints (Bogduk, 1983). Little attention has been given to studying the specific innervation pattern of the lumbar muscles and in particular LMT.

Recognizing that discrepancies existed in regards to the innervation of this muscle in the literature and many standard anatomical textbooks, Macintosh et al (1986) investigated its morphology and innervation pattern. Macintosh et al (1986) examined eight cadavers using microdissection of each medial branch from L1 to L5. Removal of sequential multifidus muscle fibers exposed deeper portions of each nerve and their course was

30 subsequently recorded. Individual branches were traced as far as possible through the muscle. Any intercommunicating branches were also recorded. The medial branch of each posterior ramus was found to cross the vertebral lamina deep to lumbar multifidus before dividing into several branches. One branch passed medially to supply the interspinous ligament, while the remaining branches supply separate muscle fascicles of lumbar multifidus. Interestingly and clinically relevant was the finding that “each medial branch innervates only those fascicles that arise from the spinous process or lamina of the vertebra with the same segmental number as the nerve. Conversely, this relationship can be expressed as, “the fascicles arising from a given vertebra are innervated by the nerve that issuesbelow that vertebra” (Macintosh et al., 1986).

Despite their pioneering work and subsequent evidence implicating LMT in the precise control of intersegmental movement, little has been done since to further unravel the detailed intramuscular nerve innervation pattern of LMT.

2.3.3.1 Motor Control of Lumbar Stability

The oseoligamentous spine is inherently unstable and has been shown to buckle under minimal loading conditions (Crisco & Panjabi, 1991). Panjabi (1992a) demonstrated the vital role of muscle to the control of spinal stability in addition to its passive elements. Although the trunk muscles are capable of providing ample stability to the spine under most day-to-day activities, control over these muscles is a function of the central nervous system (CNS) (Panjabi, 1992a). The CNS is continuously monitoring the current state of stability of the spine. It pre-plans muscle activation strategies to overcome known challenges to stability and must react quickly to unpredictable

31 challenges using information received from various sensory systems, including muscle, ligament and joint proprioceptors. To make matters even more complex, muscle activation patterns must be coordinated such that intersegmental movements such as shear and torsion between vertebrae is controlled while simultaneously allowing the spine to move.

The two primary strategies used by the CNS to control movement and stability of the trunk are: feedforward (open-loop) strategies and feedback (closed-loop) strategies.

Feedforward strategies are implemented when the CNS can extrapolate the outcome of a predictable perturbation and responds appropriately using pre-planned motor strategies.

Movements which are considered representative of feedforward movements include ballistic or repetitive movements as well as predictable challenges to spinal stability such as lifting an arm or leg away from the body. Feedback strategies cannot be pre-planned.

Movements patterns included in this category are generated and modulated by sensory inputs from various sources including the visual, vestibular and proprioceptive systems.

In reality, control over trunk stability likely occurs as the result of interplay between both these control systems.

2.3.4 Fiber Typing of LMT

The muscles of the backhave been shown to counter-act the force of gravity with nearly continuousactivity in the erect human spine (Asmussen & Klausen, 1962). To determine if Type I muscle fibers predominate in the back muscles several histochemical studies have been conducted.

2.3.4.1 Cadaveric Investigations

Previous studies using autopsy specimens have found that the LMT as well as the lumbar and thoracic components of the erector spinae muscles have a higher percentage of Type I muscle fibers compared with Type II muscle fibers (Johnson et al., 1973; Fidler et al., 1975; Jowett et al., 1975; Sirca & Kostevc, 1985; Thorstensson & Carlson, 1987;

Jorgensen et al., 1993; Rantanen et al., 1993; Mannion, Dumas et al., 1997; Mannion,

Weber et al., 1997). Furthermore, the proportion of Type I fibers in the thoracic erector spinae muscles has been reported to be as high as 75% (Sirca & Kostevc, 1985). A lower percentage of Type I fibers has been found in the lumbar erector spinae muscles, with reported percentages varying from 58-67% (Fidler et al., 1975; Jorgensen et al., 1993;

Mattila et al., 1986; Sirca and Kostevc, 1985).

Relatively few studies have compared the composition of the LMT to the lumbar erector spinae muscles. Those that do are insufficient in one or more respects. Sirca and

Kostevc (1985) reported a higher percentage of Type I fibers in the LMT (63%) compared to the lumbar longissimus (57%). Although a relatively large number of autopsy specimens were used (21 male subjects), muscle biopsies were harvested at only one site, that is, at the level of the second lumbar spinous process. Similarly, Verbout et al (1989) reported 13% more slow twitch fibers in medial column muscles (i.e. LMT) than in lateral column muscles (i.e. the lumbar erector spinae). Once again although a large number of specimens were used (30 cadavers), only a small number of muscle biopsies were taken from the lumbar column per specimen. In addition, the precise location and level of the spine from which these muscle biopsies were harvested is not clear in this study.

In contrast to these previous findings, Jorgensen et al (1993) evaluated the histochemical composition of the LMT and lumbar erector spinae musclesin six male cadavers with no history of LBP (age 17-29). Superficial and central portions of LMT, longissimus, and iliocostalis were sampled bilaterally at the level of L3 vertebra. When the results for superficial and central portions were pooled, it was found that longissimus had a significantly greater percentage of Type I fibers (70.5%) compared with LMT

(54%) and iliocostalis (55%). Similar to the study by Sirca and Kostevc (1985), the study by Jorgensen et al (1993) is limited in that these authors sampled paraspinal muscle tissue from only one level within the lumbar column.

In summary, there is conflicting data on the fiber type composition between LMT and the erector spinae muscles. The studies described above use a relatively large number of specimens but have arrived at their conclusions using muscle biopsies taken from only a few sites within each specimen. Hence, one cannot say for certain if or how the distribution of Type I and II fibers changes throughout the extent or “volume” of each of these paraspinal muscles.

2.3.4.2 In Vivo Investigation

Thorstensson and Carlson (1987) studied the distribution of Type I and II fibers in

9 male and 7 female subjects with no history of LBP aged 20-30 years. Muscle samples were taken from superficial longissimus and LMT on the left side of the L3 spinal level.

Although a higher percentage of Type I fibers was present in both, no significant difference in the relative number of fiber types between LMT (62%) and longissimus

(57%) were observed. The authors concluded that these results did not support the

34 hypothesis of functional differentiation between these two anatomically different regions of the lumbar erector spinae.

Mannion, Weber et al (1997) obtained superficial samples of LMT at the level of

L3 or L4 during spinal surgery from 21 patients with low back pain and from the belly of the lateral tract (iliocostalis/longissimus muscles) in 21 control volunteers matched for gender, age and body mass. The biopsies from each group were subjected to routine histochemical analysis to determine characteristics of muscle fiber types. Their results showed that the proportion of Type I (51%-66%) versus Type II (7.9%-24.4%) fibers was significantly higher in both the patient and the control groups. In addition Mannion,

Weber et al (1997) were able to demonstrate that the proportion of Type I fibers was significantly higher in the muscles of the controls than in patients with LBP (p=0.0001).

Since muscle biopsies taken from LMT in this study were those belonging to patients with LBP, one cannot draw any conclusions on the possible percentages of Type I/II fibers in normal individuals. Indeed much of the in vivo data on the fiber type composition of LMT has been drawn from studies on patients with a history of LBP or spinal pathology (Mattila et al., 1986; Meier et al., 1997; Bajek et al., 2000) that generally report a high percentage of Type I muscle fibers.

2.3.4.3 Comparison of Cadaveric and In Vivo Measurements

Very few studies compare cadaveric measures to in vivo measures of muscle fiber type, diameter and volume/CSA. In so far as changes in muscle fiber diameters are concerned, it has been found that fiber diameters decrease when skeletal muscle enters a state of rigor mortis (Levine & Hegarty, 1977). This decrease in fiber diameters is

35 attributed to the movement of cytoplasmic fluid into the extraxellular space. The degree to which this occurs however has not been quantified. Hence, direct comparison between cadaveric and in vivo muscle fiber diameters is not possible at this time. One factor complicating such a comparison is that most in vivo muscle biopsies are routinely fixated with different chemical agents which in turn can also influence the diameter and size/volume of muscle fibers. In one study by Stickland (1975), different fixatives were demonstrated to decrease the diameters and volumes of muscle tissue by as much as

62%-83% and 24%-44% respectively.

Unfortunately, there are no studies which have examined differences in fiber type distribution between cadaveric and in vivo skeletal muscle tissue making comparison between these studies impossible.

2.4 Functions of LMT based of Morphological, Biomechanical, Electromyographic and Clinical Evidence

The structure of LMT is extremely complex allowing it to assume several possible roles. Of these roles, considerable evidence exists to support its’ proposed function in supporting and stabilizing the lumbar spine.

2.4.1 Morphological Evidence

Based solely on morphological evidence, LMT is likely to provide lumbar spinal stability. Its fiber bundles have not one, but multiple points of attachments onto the lumbar spine, sacrum and pelvis, thereby making it possible to control the intersegmental movement that occurs between adjacent vertebrae (Macintosh & Bogduk, 1986; Crisco &

Panjabi, 1991). The segmental arrangement of its fascicles coupled with its segmental

36 innervation makes LMT well suited for the precise control of movement in the lumbar spine. However, because of its overall small size as compared with other trunk muscles and its proximity to the centre of rotation of the intervertebral joints, the LMT muscle has little mechanical effect. Nevertheless, the arrangement of its fascicles suggests that LMT can also produce some movement in the sagittal plane. A posterior view of the muscle shows that its fascicles are aligned primarily in a vertical plane with a slight horizontal deviation (Figure 2.8). Hence, LMT is considered to be an extensor muscle along with the erector spinae muscles of the lumbar spine (Macintosh & Bogduk, 1986). Only slight movement is possible in the horizontal plane making LMT a poor rotator of the lumbar spine (Macintosh & Bogduk, 1986).

Because of its polysegmental nature, LMT also exerts indirect effects on interposed vertebrae. Since the line of action of the fiber bundles of LMT lie behind the lumbar lordosis, contraction of LMT would result in something of a “bowstring” effect on vertebrae situated between those segments to which LMT attaches (Adams, 2002).

This bowstring effect would result in compression of posterior column structures and tensioning of anterior column structures. Hence, a secondary effect of LMT on the lumbar spine might be to maintain and accentuate the lumbar lordosis.

Lastly, the deepest, laminar fibers of LMT attach onto the joint capsules of each zygapophyseal joint and hence contraction prevents entrapment of soft tissue structures during spinal movements, particularly extension (Bogduk & Endres, 2005).

In summary, based primarily on morphological data, LMT is more likely to function as a stabilizer of the lumbar spine than as a prime mover. In many ways, the

LMT is to the lumbar spine as the rotator cuff muscles are to the shoulder joint.

Fig. 2.8. Posterior view of lumbo-sacral spine showing typical orientation of fascicle of LMT. The line of action of each fascicle of LMT (bold arrow) can be resolved into a major vertical vector (V) and a smaller horizontal vector (H).

2.4.2 Biomechanical Evidence

Biomechanical studies have provided some of the most convincing evidence of the roles of LMT. The majority of these investigations, especially those referring to the deep laminar fibers of LMT, are linked to its primary stabilizing function. Biomechanical models of LMT function have confirmed the extensive role played by this muscle in producing torque and augmenting spinal stability.

2.4.2.1 Role of LMT in Torque Production

Although two or more muscles may perform similar functions, it has been proposed that certain muscles may exhibit functional differentiation. Bergmark (1989) suggested that the trunk muscles could be divided into one of two categories: ‘global’ or

‘local’ muscles. Global muscles cross several segments (i.e. vertebrae) and have a large

38 moment arm which is capable of generating substantial torque. Local muscles on the other hand cross only one or at most a few vertebral segments. They have a relatively small moment arm and are better suited to maintain joint and hence spinal stability.

Bergmark (1989) proposed that the muscles belonging to the global muscle system (e.g. thoracic divisions of the erector spinae muscles) are the main torque producing muscles of the lumbar spine. In contrast, muscles such as the LMT are included in the local muscle system. These later muscles generally lie deeper, attach directly to the lumbar vertebrae, and function primarily to stabilize the lumbar motion segments. While this classification is useful, it should not be used dogmatically, particularly when dealing with complex muscles such as LMT. Hence, although LMT is apparently well designed to provide stability of the lumbar vertebrae, it may also play a part in torque production.

The fascicles of LMT run obliquely when viewed from its posterior aspect.

Hence, their line of action can be divided into two vectors (Figure 2.8), a large vertical vector and a smaller horizontal vector (Macintosh et al., 1986). Since the magnitude of the vertical vector greatly exceeds that of the horizontal vector, the principle action of

LMT is lumbar extension. LMT thus contributes to the total extensor moment and is estimated to provide approximately 20% to the overall extensor torque at the L4 and L5 vertebral levels (Bogduk et al., 1992). Even though LMT has the distinction of being the largest muscle at the lumbosacral junction, it is at a mechanical disadvantage in producing extension of the thoracic cage on the pelvis. The majority of the torque is produced by the biomechanically well-positioned thoracic components of the erector spinae muscles with their long lever arm (Bogduk et al., 1992).

2.4.2.2 Role of LMT in Spinal Stability

The stabilizing role of LMT has been elucidated largely through biomechanical investigations addressing its possible roles in maintaining the lumbar lordosis and controlling shear forces.

2.4.2.2.1 Role of LMT in Maintaining the Lumbar Lordosis

LMT consists of many fiber bundles arranged into fascicles, the longest of which can accentuate the lumbar lordosis (Bogduk & Endres, 2005). In addition, the fact that each band of LMT has its own nerve supply (Section 2.3.3) implies that each vertebra is independently controlled and therefore “the curvature of the lumbar spine can be adjusted very precisely to match the loading being imposed. It is the overall curvature, or posture, of the spine which determines its stability” (Aspden, 1992).LMT can anteriorly tilt the pelvis relative to the lower lumbar segments (Bogduk & Endres, 2005). This allows LMT to maintain control over the lumbar lordosis while the erector spinae simultaneously moves the trunk, which may offer the spine some safeguard from injury, particularly during lifting activities.

Aspden (1992) suggested another mechanism whereby LMT may enhance lumbar stability by controlling the lumbar lordosis. Contraction of LMT increases the lumbar lordosis, thereby generating compressive forces on the interposed vertebrae. Compression increases the resistance of the lumbar spine to torsional forces thereby augmenting lumbar stability.

2.4.2.2.2 Role of LMT in Controlling Shear Forces

Anterior/posterior shear relates to the movement of one vertebra forward/backwards relative to the sub-adjacent vertebra. Bogduk et al (1992) found that contraction of LMT led to posterior shear forces at the L1 to L4 vertebral levels, and anterior shear forces at the L5 level. To explain the generation of these forces,the authors proposed that shear is generated not only by muscle fiber bundles attached to their vertebrae of origin, but also on all interposed vertebrae between their origins and insertions. For this reason, shear forces generated by lumbar fiber bundles reach a maximum at intermediate segments. Secondly, although each fiber bundle or fascicle of LMT may contribute minimally to shear forces on lumbar vertebrae in the upright position, when their individual contributions are summated the resultant shear force generated is substantial.

LMT is also theorized of being capable of controlling shear forces in standing and forward flexion. Cholewicki et al (1991) proposed that LMT opposes the anterior shear caused by bending and lifting due to production of posterior shear by LMT.

2.4.2.2.3 Biomechanical Models for the Stability Role

It has been proposed that the spinal muscles may impart stabilization via control of the spinal segments’ neutral zone (NZ) (Panjabi 1992b). The neutral zone is a region of intervertebral motion around the neutral posture where little resistance is offered by the passive spinal column. The NZ appears to be a clinically important measure of spinal stability function. Its size may increase with injury to the spinal column, which in turn may result in spinal instability or low-back pain. Panjabi et al (1989) looked at the effect of intersegmental muscle forces on spinal instability in an in vitro study

41 performed on intact and sequentially injured fresh lumbar spinal segments. Simulated muscle forces representing the intersegmental LMT, interspinales and rotatores muscles were applied to the specimens. Panjabi et al (1989) concluded that the “intersegmental nature of the deep muscle group gives a tremendous advantage to the neuromuscular control system for fine tuning the stability of the spine”. LMT fibers are placed close to the centers of rotation of spinal movements and connect adjacent vertebrae at appropriate angles.

In a three dimensional study of the lumbar spine mechanics, McGill (1991) provided supporting evidence of the stabilizing role of LMT. Based on the results of this study, McGill (1991) concluded that the “unchanging geometry of the LMT through a range of postures” indicates that “the purpose of LMT was to finely adjust vertebrae with small movements rather than to function as a prime mover. The results of this study showed that LMT could function in this way at any physiological posture”.

Goel et al (1993) provided further evidence tosupport the stabilizing function of

LMT using a combined finite element and optimization approachto study the effects of muscles on the of biomechanicsof the lumbar spine. Briefly, two finite element models

(ligament and muscle) of an L3-4 motion segment were created and subsequently compared. Muscles included in the muscle model were the interspinous and intertransverse muscles, LMT and the quadratus lumborum. The results of this study showed that the incorporationof muscular forces led to a decrease in anteroposterior translation and flexion rotation (displacements in the sagittal plane)of the L3-4 motion segment. Hence, muscles imparted stability on the ligamentous system. The addition of muscles also led to a decrease in stresses in the vertebral body and intervertebral disc.

However, the load bearing of thefacet joints increased, indicating that these joints play a significant role in “transmitting loads in a normal intact spine” (Goel et al., 1993). This study supported the earlier findings of Farfan (1975). Muscle dysfunction (simulated by decreasing the computed force in the muscles to 90%, 80% and 70% of the original values) destabilised the motion segment. This led to a shift of loads to the disc and ligaments and decreased the role of the facet joints in transmitting loads (Goel & Gilbertson,

1995).

The stabilizing function of the LMT was also demonstrated in a biomechanical in vitro study performed by Wilke et al (1995). This study investigated the influence of five different muscle groups on the monosegmental motion of the L4-L5 segment during the movements of flexion/extension, lateral flexion and axial rotation. The muscles examined were the LMT, lumbar erector spinae and psoas major. Seven human lumbosacral spines were tested on a spine tester that allowed simulation of muscle forces. The combined muscle action of the muscles tested was found to decrease the total range of motion and neutral zone motion of the L4-L5 segment. Total range of motion was decreased by 93% in flexion, and 85 % in extension. The total neutral zone motion in flexion and extension was decreased by 83 %.

This supported the findings of Steffen et al (1994), who also used a biomechanical in vitro experimental design for assessing lumbar instability, and found that the influence of the LMT decreased the neutral zone in flexion and extension. In lateral flexion, Wilke et al (1995) showed that the total range of motion was decreased by 55 % and the neutral zone motion was decreased by 76%. In axial rotation, total range of motion was decreased by 35%, but the neutral zone did not change significantly.

The studies of Crisco and Panjabi (1991), Goel and Gilbertson (1995), Panjabi et al (1989), Steffen et al (1994) and Wilke et al (1995) exemplify the stabilizing role of the

LMT on lumbar spinal segment in the sagittal and frontal planes. Unfortunately, the role of the LMT in stabilizing axial rotation is not as clearly defined (Wilke et al., 1995).

By modeling the actions of LMT and the lumbar erector spinae, Macintosh et al

(1993) concluded these muscles were incapable of providing axial stability of the lumbar spine due to insufficient torque production by these muscles in trunk rotation. These authors however, did not consider that muscle stiffness, which is a critical element in joint stabilization, can be produced at low levels of maximal voluntary contraction

(MVC) (Hoffer & Andreassen, 1981).

The role of the LMT in axial rotation has also been considered in relation to the abdominal muscles. Joint stiffness can be improved by co-contraction of agonist and antagonist muscles (Lee et al., 2006). Therefore, although LMT may not be a strong torque producer in rotation, this does not preclude it from contracting in association with other muscles such as the oblique abdominals in stabilizing trunk movement in axial rotation (Bogduk & Endres, 2005).

A study by Kaigle et al (1995) took the research of the previous authors [Crisco and Panjabi (1991), Goel et al (1993), Goel and Gilbertson (1995), Panjabi et al (1989),

Steffen et al (1994) and Wilke et al (1995)] one step further. While these researchers have used biomechanical in vitro experimental designs, Kaigle et al (1995) developed an in vivo animal model of lumbar segmental instability. In this investigation, passive stabilizing structures (disc, facet joints and ligaments) were transected, and the effects of active musculature on spinal kinematics were examined in 33 pigs. Muscles surrounding

44 the spine (LMT, lumbar erector spinae, quadratus lumborum and psoas major and minor) were subjected to muscle stimulation using wire electrodes. Kaigle et al (1995) showed that increased combined muscular activation stabilized the injured motion segment by reducing aberrant patterns of motion in the neutral zone. However, these results must be interpreted with caution as experiments on quadrupeds may not be directly applicable to bipeds because of the different directions of external forces relative to the structures of the vertebral column. Overall, Kaigle et al (1995) concluded that rehabilitation for patients with segmental spinal instability should focus on the stabilizing influence of the surrounding spinal musculature.

2.4.2.2.4 Role in Providing Stiffness to the Spine

LMT has been shown to decrease the available movement of the lumbar spine and neutral zone motion of individual lumbar motion segments, thereby stiffening the spine

(Wilke et al., 1995). Stiffness is defined as resistance to deformation. If a structure has increased stiffness, greater outside force would be required to deform the structure to the same amount (Porterfield & DeRosa, 1998), and consequently, it is considered more stable.

Muscle stiffness is the ratio of force change to length change and consists of two components: reflex mediated muscle stiffness and intrinsic muscle stiffness (Johansson et al., 1991). The neuromuscular system regulates muscle stiffness in postural control

(Crisco and Panjabi 1991), and the stiffness of a muscle can be increased by increasing the neural outflow to it and thus increasing muscle tone (Porterfield & DeRosa, 1998).

This neural outflow affects both the alpha and gamma motor systems. Peripheral

45 feedback from the joint and ligament afferents may also regulate muscle stiffness via effects on the gamma spindle system (Johansson et al., 1991). The gamma system sensitizes the spindle to movement (Porterfield & DeRosa, 1998).

The bending stiffness of the spine will also be influenced byother factors such as the thoracolumbar fascia (Porterfield & DeRosa, 1998), whichlimits the radial expansion of the backmuscles (Aspden, 1992). It has been sugested that contraction of the back muscles, including the LMT and the lumbar erector spinae muscles, exerts a pushing force on the fascia (Farfan, 1973). The influence of the LMT and the lumbar erector spinae muscles on the thoracolumbar fascia was investigated by Gracovetsky et at (1977) using a mathematical model. It was proposed that since the thoracolumbar fascia surrounded the back muscles, it could serve to brace these muscles. The authors called this the hydraulic amplifier mechanism. These forces may result in increased lumbar spine stiffness and contribute to lumbar stabilization.

2.4.3 Electromyographic Evidence

Electromyographic analysis has allowed evaluation of the functionof the LMT through determination of invivo muscle activation. Many classic studies have been performed using indwelling electrodes. A tonic or almost continuous level of activation of the LMT has been demonstrated in many of these studies, which have examined the role of the LMT in upright postures and during active movements.

2.4.3.1 LMT Activity Involved in the Maintenance of Posture

LMT has been shown to be continuously active in upright postures, compared with relaxed recumbent positions. Along with the other paravertebral muscles, LMT provides antigravity support to the spine with essentially continuous activity (Asmussen and Klausen 1962). In fact, the LMT is probably active in all antigravity activity

(Donisch & Basmajian, 1972; Valencia & Munro, 1985).

In the standing position, slight to moderate EMG activity has been demonstrated in LMT (Donisch and Basmajian 1972, Valencia and Munro 1985). The explanation for this activity lies in the location of the line of gravity in relation to the lumbar spine

(Valencia & Munro, 1985). In approximately 75 % of people, the line of gravity passes in front of the centre of the L4 vertebra (Asmussen & Klausen, 1962). Gravity creates a force ventral to the spine which tends to pull the thorax and lumbar spine into flexion.

The LMT, which is a posterior sagittal rotator of the lumbar spine, is therefore constantly active when standing to maintain an upright posture and opposes the tendency to flexion

(Bogduk & Endres, 2005). This is an example of the tonic postural role of the LMT muscle. During walking the LMT is tonically active (Morris et al., 1962).

2.4.3.2 LMT Activity in Active Lumbar Movements

Activation of the LMT has been examined using EMG during several movements including forward flexion and extension from the flexed position, trunk extension in the prone position and trunk rotation. In all cases the function of LMT appears to be primarily one of stabilization.

As the spine bends forward from the erect standing posture, there is an increase in

LMT EMG activity (Morris et al., 1962; Valencia & Munro, 1985). At a certain point during flexion known as the ‘critical point’ the activity of the back muscles ceases (Morris et al.,

1962; Kippers & Parker, 1984). Kippers and Parker (1984) demonstrated that the EMG activity of the erector spinae ceased at about 90% of lumbar spine flexion. The critical point for LMT is not as characteristic a feature as it is for the erector spinae muscles; however a decrease in activity is evident, with EMG silence of the LMT occurring infrequently

(Valencia & Munro, 1985). Valencia and Munro (1985) proposed on the basis of these results that the LMT has a stabilizing role in flexion, providing localized control of lumbar vertebrae motion during this activity. An alternative explanation could relate to the small length changes in the LMT related to sagittal movements (McGill, 1991).

Extension of the trunk from the flexed position predictably evokes high levels of

LMT activity (Morris et al., 1962; Donisch & Basmajian, 1972). Significant activity of the

LMT also occurs when the trunk is extended or hyperextended in the prone position

(Jonsson, 1970; Donisch & Basmajian, 1972; Valencia & Munro, 1985). Valencia and

Munro (Valencia & Munro, 1985) also found that unilateral hip extension in the prone position evoked high levels of LMT activity on both sides of the spine. During these highly loaded activities, similar to extension of the trunk from a flexed position in standing, LMT activity causes posterior rotation of the lumbar vertebrae (Bogduk & Endres, 2005), and controls the lumbar lordosis (Aspden, 1992). Although activity in LMT is marked in extension, the majority of the actual trunk extension torque (80% at the L4 and L5 vertebral levels) is provided by the thoracic components of the erector spinae muscles (Bogduk et al.,

1992).

During trunk rotation, the LMT has been shown to be active bilaterally in both

48 ipsilateral and contralateral rotation of the trunk in sitting and standing (Morris et al.,

1962; Jonsson, 1970; Donisch & Basmajian, 1972). For this reason, it has been suggested that in rotation, the LMT acts as a stabilizer rather than a prime mover

(Valencia & Munro, 1985).

2.4.3.3 LMT Activity During Internal and External Perturbations of the Trunk

Studies used to elucidate the role of LMT in the control of movement and/or stability have evaluated the recruitment of this muscle when spine stability is challenged due to both predictable and unpredictable challenges to trunk stability (Moseley et al.,

2002, 2003). Electromyographic activity was recorded from the superficial and deep fibers of LMT, erector spinae, transversus abdominis, and deltoid muscles using fine wire electrodes during single and repetitive arm movements. The activation patterns of these muscles in response to elevating the arm were compared to one another. During single arm movements, the onset of electromyographic activity in deep fibers of LMT was shown to occur irregardless of the direction of arm movement while the superficial fibers of LMT were activated in a direction specific manner. These researchers postulated that the deep fibers of LMT controlling intersegmental motion while the superficial fibers control spinal orientation (Moseley et al., 2002). To explore the activation pattern of

LMT when the spine cannot make any predictions to stability and hence pre-plan a motor response, Moseley et al (2003) recorded the electromyographic activity from the superficial and deep fibers of LMT as a result of unexpectedly dropping a weight into a bucket that was held by a standing subject with the eyes closed. These researchers did not observe differential electromyographic activity in the deep and superficial fibers of

LMT (Moseley et al., 2003). Hence, it was suggested that the deep fibers of LMT may help to stabilize the lumbar spine but only when the CNS can anticipate the timing of the perturbation.

2.4.4 Clinical Evidence

Therapeutic exercise programs are an important component in the treatment and management of clients with LBP. These exercise programs are directed towards recruiting one or a combination of different trunk muscles with the ultimate goal being to improve

“stability” of the lumbar spine. In doing so it is generally felt that both a reduction in symptoms and reoccurrence rates of LBP would naturally ensue. Despite this, there is controversy over which muscles are most important in attaining optimum function and performance following or during an episode of LBP. While some authors favor activation of the paraspinal muscle group as a whole to achieve control over spinal motion (Porterfield &

DeRosa, 1991, 1998; McGill, 2001), others (Richardson et al., 2004) propose the initial activation and rehabilitation of specific muscles which are preferentially suited to stabilizing the lumbar spine. For the later group, specific exercises have been designed to target LMT in such a precise manner that normal function and hence stability to the spine would naturally be restored (O'Sullivan et al., 1997; Richardson, 1999; Richardson et al., 2004). Hides et al

(2001) demonstrated that by using specific exercise therapy to activate LMT it was possible to dramatically reduce the reoccurrence rate of acute LBP while O’Sullivan et al (O'Sullivan et al., 1997) provided convincing evidence of reduced pain and disability in patients with chronic LBP. Furthermore, patients who suffered from poor outcome following back surgery, also exhibited local denervation of LMT (Zoidl et al., 2003).

2.4.5 Summary

Despite evidence to support the role of LMT in maintaining and facilitating movement, providing stability, controlling shear forces and maintaining the lumbar lordosis, there is lack of consensus as to how LMT is capable of supporting these multidimensional task oriented roles within the in vivo lumbar spine. Even though the morphology/anatomy of this muscle has been looked at previously, it has not provided the necessary substrate data with which to ‘piece together’ and thereby formulate a functional paradigm of this muscle.

What is needed is a robust, anatomically accurate, 3D representation of LMT which incorporates all of its fiber bundles, nerve supply and fiber type distribution throughout its volume! Architectural data, innervation patterns and fiber type constitution may then be explored and analyzed to draw meaningful conclusions as to the relationship between structure and function. In the past the ability to render such a complete model of LMT has been impossible, however, with the advent of novel microdissection and digitization techniques pioneered predominately in this laboratory a virtual model of the entire back musculature is now possible.

Chapter 3 Hypothesis and Objectives

3.1 Hypotheses

The human lumbar multifidus is functionally subdivided based on regional differences in muscle architecture, innervation pattern, and fiber type distribution.

3.2 Objectives

The objectives of this study are as follows:

1. (a) To serially dissect and digitize lumbar multifidus in situ throughout its entire

volume from L1 to L5;

(b) To construct a 3-D model of each specimen using MAYA™ (Alias Systems

Corporation, Toronto, ON) in conjunction with additional software developed in

the laboratory.

(c)To model and quantify the architecture of LMT throughout its volume.

2. To determine if the lumbar multifidus is architecturally divided based on gross

morphology and architectural parameters (fiber bundle length, fiber bundle angle,

and volume).

3. Digitize and model the innervation of lumbar multifidus in 3D throughout its

volume.

4. Catalogue the distribution of Type I and Type II muscle fibers throughout the

volume of lumbar multifidus from L1 to L5.

5. Use the architectural, nerve and fiber type data to establish a functional paradigm

which explains the role of lumbar multifidus within the lumbar spine.

51 52

3.3 Significance

This study provides a unique three-dimensional model of the detailed architecture, innervation, and fiber type distribution of lumbar multifidus. The results will provide new insights into the architectural infrastructure of lumbar multifidus and assist in clarifying whether different regions serve different functions. It is known that lumbar multifidus is innervated segmentally from the medial branch of the posterior ramus, but what is not known is whether the medial branch divides further to supply architecturally distinct regions. An architecturally distinct region which is independently innervated is termed a neuromuscular compartment which may have functional or task-oriented roles; that is, different portions of the same muscle may be utilized proportionally depending on the task demands of the situation (English et al., 1993).

The fiber type distribution, innervation pattern and fiber architecture within each compartment could be used to predict functional properties and to elucidate how compartments coordinate with each other during movement.A functional paradigm can then be formulated which allows analysis of normal and abnormal movement, which is the basis for understanding and managing dysfunctions of the musculoskeletal system

(English et al., 1993)

Chapter 4 Methods

The methods are divided into three sections. Section 4.1 outlines the procedure used to dissect, digitize, model and quantify the architectural parameters of LMT. A flowchart which outlines this process is provided in Figure 4.1. In section 4.2 the method used to delineate the nerve innervation pattern of LMT throughout its volume is discussed. The methods used to determine the proportion of Type I/II fibers throughout the volume of LMT as well as the least mean diameter (LMD) for each fiber Type is described in section 4.3.

4.1 Digitization, modeling and quantification of architectural parameters of LMT

4.1.1 Specimens

Ten formalin embalmed human cadaveric specimens (9M/1F) with a mean age of

80±11 years were studied. Specimens with visible evidence of musculoskeletal deformity, muscle pathology, indication of previous surgery or trauma were excluded.

Ethics approval was received from the University of Toronto Research Ethics Board

(Protocol Reference # 20830).

4.1.2 Serial dissection and digitization of muscle fiber bundles, tendons and spinal column

The LMT was exposed by removing the skin, fascia, superficial muscles and aponeuroses. Each specimen was securely bound to a metal tray and three reference points (bilateral posterior superior iliac spines, and sacral apex) were demarcated clearly by drilling in screws. These reference points were necessary in order to later reconstruct

53 54

Figure 4.1 Digitization of human LMT.

55 the 3D model of LMT from the digitized data. Using a 1.75x magnifier, muscle fiber bundles of LMT were identified and traced along their full length craniocaudally starting at the L1 spinous process. The course of each fiber bundle was delineated by marking its superior and inferior attachment sites and 5-20 intervening points using a fine paint pen (Figure 4.2).

Figure 4.2 Delineation of muscle fiber bundle of LMT (left, lateral view of lumbosacral spine). Small black markings (indicted with arrows) represent points used to outline the course of a muscle fiber bundle. Points were subsequently digitized. Tendons were digitized in a manner similar to muscle fiber bundles (small white markings on specimen).

The x, y, and z coordinates of each point were then obtained using a Microscribe®

G2 Digitizer (Immersion Corporation, San Jose, CA, USA). The removal of individual fiber bundles permitted the identification, marking, and digitization of successively deeper segments of the muscle. The dissection and digitization process continued

56 sequentially from cranial to caudal and from superficial to deep until the entire LMT had been resected from the lumbar spine, ilium and sacrum.

During the dissection process, the surfaces of all tendons forming part of the structure of LMT were demarcated using small points. Points were placed on the surface of each tendon from its superior to inferior ends across its entire width/girth. These points were then digitized in a linear and sequential manner (Figure 4.2 and 4.3).

Figure 4.3. Close up, lateral view of LMT originating from the L1 spinous process showing a small segment of tendon (left, lateral view of spine). The tendon was marked using small points (black dots) which were digitized in a sequential manner (e.g the points in row 1 followed by those in row 2, etc…) from superior to inferior. Solid blue arrows demonstrate direction of digitization procedure.

Lastly, the surface and peripheral outline of each lumbar vertebra, the sacrum and ilium were digitized using a grid that was marked or etched directly onto each of these structures using a fine marker (Figure 4.4).

Figure 4.4. Right, lateral view of digitized lumbar spine and sacrum as viewed in Autodesk® Maya®. Each small red dot represents a digitized point.

4.1.3 Microscribe® 3G2 Digitizer

The Microscribe® 3G2 Digitizer consists of a small, sensorized mechanical arm that sits on the support base. The probe has six joints (joints 0-5), as illustrated in Figure

4.5.

Figure 4.5 The Immersion Company Microscopic 3-G2 Digitizer.

Each rotary joint represents one degree of freedom, and thus the probe has six degrees of freedom, allowing simultaneous positioning and orientation of its tip. A counterbalance is placed close to the base to minimize the user’s fatigue. The tip position relative to the base is obtained through direct kinematics calculations, based on sensor values and the length of the links. Software on the host computer reads the joint sensors and uses its kinematic model to determine where the tip is. A binary switch on a footpedal or hand switch is used to select/deselect virtual objects, navigate, or mark points on the real objects surface for digitization purposes.

Calibration is done by placing the probe tip in a housing close to the base and reading the rotary sensor data. These readings are then compared with the stored values corresponding to the joint angles when the probe arm is in this known position. The G2 model, which is the digitizer used in this study uses high-resolution joint sensors with a tip accuracy of 0.23 mm (0.009 in.).

4.1.4 3D reconstruction and modeling of LMT

The digitized data were used to reconstruct the structure of the LMT as it appeared in situ using Alias® Maya® in conjunction with customized software developed in our laboratory. The virtual model created for each specimen was then used to visualize and document the attachment sites, spatial orientation and distribution of fiber bundles and tendons in relation to the bony structures. Morphological differences observed within the LMT were used to demarcate architecturally distinct regions.

4.1.5 Quantification of architectural parameters of LMT

The architectural parameters of LMT investigated in this study were FBL, extramuscular tendon length, FBA, and volume.

4.1.5.1 Fiber Bundle Length (FBL)

FBL (cm) was calculated as the sum of the distances between each of the digitized points along the length of a fiber bundle from superior to inferior. For example, Figure

4.6 demonstrates the length of a fiber bundle measured from its superior attachment to the tip of the L1 spinous process to its inferior attachment to the S1 mammillary process.

Figure 4.6. Measurement of muscle fiber bundle length and tendon length of LMT (right, lateral view of lumbosacral spine). Length of muscle fiber bundle (red line) delineated using marked points (yellow dots) Distances between marked points were summed to yield the total length for a particular fiber bundle. Fiber bundle length (FBL); Tendon length (TL); Lumbar multifidus (LMT); Lumbar (L).

4.1.5.2 Tendon Length

As described in section 4.1.2, the span of each tendon associated with LMT was recorded using the digitizer. The extra-muscular tendon length (Figure 4.6) was then

60 computed using a plug-in algorithm developed in our laboratory for Alias® Maya®. The algorithm calculated the length of each line traced using the digitizer by summing the distances between digitized points from superior to inferior (Figure 4.3). Average tendon length was derived by summing the lengths of each line and dividing by the total number of lines digitized.

4.1.5.3 Fiber bundle angle (FBA)

Similar to FBL and tendon length, FBA was computed using a plug-in algorithm developed for Alias® Maya®. FBA was defined as the angle formed in the sagittal plane between a tangent line drawn through the centre of each lumbar spinous process (L1-L5) and the fiber bundles attaching to these vertebrae (see Figure 4.7).

Figure 4.7. Calculation of muscle fiber bundle angle (right, lateral view of lumbosacral spine). Each fiber bundle produces a unique FBA with respect to its vertebrae of origin. Each astrics (*) therefore represents a FBA calculated using a customized plugin in Autodesk® Maya®. Fiber bundle angle (FBA); Lumbar (L).

4.1.5.4 Volume

Inspection of the morphology and architecture of LMT showed that it consisted of several bands or fascicles. Each fascicle was comprised of several muscle fiber bundles with unique superior/inferior points of attachment onto the lumbosacral spine.

Fiber bundles constituting each fascicle of LMT were stripped away carefully.

Fat and other non-skeletal muscle tissue was removed prior to adding the tissue to a graduated cylinder containing distilled water. Once all fiber bundles constituting a given fascicle had been added the amount of water displaced was recorded. This process was repeated for each fascicle of LMT.

+ =

Muscle fiber bundle water displacement

Figure 4.8. Measurement of muscle volume.

4.1.6 Statistical analysis of architectural parameters

Fiber bundle length, fiber bundle angle, and volume of architecturally distinct parts of the muscle were characterized with descriptive statistics (mean, standard deviation). ANOVA followed by Tukey’s post-hoc test were carried out to compare means. Statistically significant difference was declared at p-value = 0.05 level with two

62 tails. K-means clustering of the architectural data was performed to determine if the data could be grouped into distinct clusters. SPSS® version 14.0 under Windows® was used for all analyses.

4.2 Digitization and modeling of the intramuscular nerve distribution of LMT

4.2.1 Specimens

Three formalin embalmed cadaveric specimens with no visible signs of lumbar muscle pathology and/or deformity were used in this part of the study. All specimens were obtained from the Division of Anatomy, University of Toronto. Ethics approval was received from the University of Toronto research ethics board (Protocol Reference

#11761).

4.2.2 Serial dissection and digitization of the medial branch of the posterior rami of L1 to L5.

The LMT was exposed unilaterally by removing the skin, fascia, superficial muscles and aponeuroses. Medial portions of the erector spinae muscles were retained to avoid cutting through any nerves during cleaning. Each specimen was secured firmly to a metal aluminum tray to prevent shifting during the digitization process and three reference points (bilateral posterior superior iliac spines, and sacral apex) were demarcated by drilling three Robertson® #2 screws directly into solid bone. The square head of each screw formed a depression into which the stylus tip of the digitizer could

"sink".

Nerves were dissected with the aid of a x40 dissecting microscope. In order to ensure preservation of the posterior ramus, its medial branch and LMT, each specimen

63 was initially dissected using an anterior approach. All abdominal viscera and musculature (e.g. psoas major) were removed and the anterior surface of the lumbar spine cleaned on order to identify the anterior rami. Anterior rami were then traced back to their origin from the spinal nerves. The posterior rami and their branches were then traced inferiorly. The medial branch of each posterior ramus was identified as it exited the intervertebral foramen and passed over the root of the transverse process. This short extramuscular portion of the nerve was digitized.

The medial branch was then traced intramuscularly through the volume of lumbar multifidus as illustrated in figure 4.9 and described below:

1. Individual muscle fiber bundles on the surface of LMT were localized,

digitized and carefully stripped away in the manner outlined in section 4.1.2.

2. Small segments of the medial branch were then exposed, marked and

digitized.

3. Serial dissection and digitization of LMT fiber bundles exposed deeper

segments of the medial branch. These small nerve segments were then

digitized.

4. Dissection and digitization continued until the intramuscular course of the

nerve and all of its branches could no longer be followed using a surgical

microscope.

Once dissection and digitization of the muscle and nerve was complete and each specimen had been denuded of all ligament and remaining soft tissue structures, the lumbosacral spine was digitized. Surface reconstruction entailed placing small dots (ie, formulating a grid) using a fine marker over the entire surface of each lumbar vertebrae,

64 the sacrum and ilium (Figure 4.10). Points on the lumbosacral spines were separated by

2-mm intervals. Each point was then digitized using the Microscribe 3G2 Digitizer to obtain x, y, and z coordinates.

Figure 4.9. Flowchart outlining the process of serial dissection and digitization of the medial branch of the posterior ramus.

Figure 4.10. Digitization of bony skeleton, 3D reconstruction. Small black dots on bony surfaces were digitized for the purpose of reconstructing the spine in 3D.

4.2.3 Reconstruction, modeling and analysis of intramuscular nerve distribution

The digitized data were used to generate a fully manipulable model of each medial branch of the posterior rami (L1 to L5) of each specimen in Autodesk® Maya®. In turn, the digitized muscle fiber bundles and osseous skeleton data were used to generate a

3D model of the musculoskeletal lumbar spine. Lastly, the 3D model of the LMT, including its bony correlates was combined with the digitized nerve and its branches.

The 3D model of the medial branch created in Autodesk® Maya® was fully manipulable, allowing its innervation pattern to be visualized and recorded from several different perspectives as it entered, divided into branches and supplied different regions of the muscle. Muscle fiber bundles and portions of the lumbosacral spine could be added to or subtracted from the model to allow the course of each nerve to be traced through the

66 volume of LMT more readily. Screen images of the medial branch and its subdivisions were taken with and without the inclusion of muscle fiber bundles and skeletal components. These images are included in the results section of this thesis.

4.3 Fiber typing of LMT

This section of the study was carried out after the results of the muscle architectural study were known.

4.3.1 Specimen(s)

One male cadaveric specimen (age 68 years) with no visible evidence of musculoskeletal deformity, pathology, or indications of previous surgery/trauma was used to obtain muscle biopsies. Ethics approval for this portion of the project was received from the Chief Coroner and General Inspector of Anatomy of the Province of

Ontario and from the University of Toronto (Protocol Reference # 21965).

A total of twenty nine muscle biopsies measuring approximately 1 cm3 were harvested from different locations (superior, middle or inferior) within the deep, intermediate and superficial regions of LMT from L1 to L5 unilaterally (Table 4.1 and

Fig. 4.11). Staining of Type I and II fibers was carried out using myosin immunohistochemistry.

Figure 4.11. Lateral view of LMT showing deep (purple), intermediate (yellow) and superficial (red) regions. Small tissue samples were taken from superior (S) and inferior (I) or middle (M) portions of the muscle depending on the region being biopsied.

Superior Division of Inferior Biopsy Number of Total number of biopsies attachment multifidus attachment Location biopsies per level (spinal level) Deep L3 Middle 1 Superior Intermediate L4 2 L1 Inferior 9 L5 Superior/inferior 2 Superficial S1 Superior/Inferior 2 Sacrum/ilium Superior/Inferior 2 Deep L4 Middle 1 Superior Intermediate L5 2 L2 Inferior 7 S1 Superior/Inferior 2 Superficial Sacrum/ilium Superior/Inferior 2 Deep L5 Middle 1 Superior Intermediate S1 2 L3 Inferior 5 Superior Superficial Sacrum/ilium 2 Inferior Deep S1 Middle 1 Superior Intermediate Sacrum/ilium 2 L4 Inferior 5 Superior Superficial Sacrum/ilium 2 Inferior Deep Sacrum/ilium Middle 1 L5 Superior 3 Superficial Sacrum/ilium 2 Inferior TOTAL 29

Table 4.1: Spatial distribution of muscle biopsies taken from specimen (n=1) used in this study.

4.3.2 Sectioning and Immunohistochemistry

Specimens were orientated in optimal cutting temperature compound embedding medium (Tissue Tek; Miles, Elkhart, IN, U.S.A.), snap-frozen in isopentane cooled by liquid nitrogen, and stored at -80°C. Serial sections, 14 um thick, were cut in a cryostat at -20°C and then reacted for anti-myosin (slow, Type I) after preincubation with Tris-

HCI (pH 9.0, 115ºC)toexpose antigens which were masked by the tissue fixation process.

Detection of Type I fibers was achieved using MACH3 (Biocare) probe followed by

MACH3 (Biocare) polymer. Specimens were washed in 0.5% BSA and then reacted with second labeled primary antibody, ALP-anti-myosin (fast, Type II) followed with 2nd substrate Vector Red (Vecter) and counterstained with Mayer’s Hemotoxylin (Sigma).

Using this protocol Myosin II stains as pink, whereas myosin I stain as dark brown

(Figure 4.12).

Figure 4.12 Typical microscopic view of the transversely sectioned LMT muscle. Type I fibers stain dark, while Type II fibers stain light.

4.3.3 Morphometric analyses of Type I/II fibers

Muscle fibers (typically 1,000-1,500 per slide) were assessed in order to identify

Type I(slow-twitch) and II (fast-twitch) fiber types. Digital images of each slide were captured at lowmagnification (10x) with an Olympus DP 70 camera (Olympus, Tokyo,

Japan) attached to a Leica DM 4500 B microscope (Leica N Plan 100 X 1.25, Leica,

Wetzlar, Germany). The camera settings and light level on the microscope were keptconstant for all of the images captured. Several images were taken from each slide in a sequential order by motorized stage to avoid selection bias until the entire slide was captured digitally. Computerized image analysis was performed by using image analysis software (Visiopharm Integratory System®, VIS, Copenhagen, Denmark).

This software is endowed with an image segmentation routinethat yields distinct optical density measures for predefinedobjects within the sections (i.e. Type I and II cells; objects are defined based on their shape and/or relativelevels of staining). Images were segmentedusing pixel classification. The segmentation routinepermitted the measurement of the amount of staining specifically associatedwith each fiber type. The Bayesian pixel classifier withinthe VIS software was initially "trained" to recognize differentcomponents in the section (i.e., Type I and II fibersand light anddark background), thus creating an algorithm that was applicableto all sections in the analysis, despite slight variations instaining intensity and imaging parameters. The classificationof different image components was thereafter used to quantify morphological parameters included: fiber diameters cross sectional area (CSA) and total fiber area measured in µm2. Fiber diameters were calculated using ten representative cells of each fiber type selected at random from each slide. The total area occupied by each fiber type was used to

70 determine the proportion of Type I and II fibers. The proportion of each fiber type by number was determined taking the total fiber area for a given fiber type and dividing this by the mean CSA calculated for each fiber type for a given biopsy slide.

4.3.4 Statistical analysis

The data were entered into the computer and a univariate distribution of each of the variables was examined for anomalies and errors were corrected.

The response variables used in the analyses were chosen to be the proportion of

Type I fiber area, the number of Type I fibers (expressed as a percentage), and the diameter of Type I and Type II fibers. The predictor variables were multifidus region

(deep, intermediate, or superficial), multifidus fascicles (deep: L1 to L5; intermediate: L1 to L4; and superficial: L1-L5, L1-S1, L1-Sa, L2-S1, L2-Sa, L3-Sa, L4-Sa and L5-Sa), lumbar Level (L1, L2, L3, L4, or L5), and biopsy location (superior, middle, or inferior).

Since the response variable is a continuous variable and the three predictor variables are discrete variables, analysis of variance is the appropriate method to analyze the relationship between these variables.

Statistical theory implies that when dealing with a response variable that is a proportion, it cannot be analyzed directly without violating the assumptions of standard analysis of variance. However, the theory indicates that a “transformed” version of the variable can be appropriately analyzed. The appropriate transformation is the arcsine of the square root of the original proportion values (Winer, 1971, p. 400).

The proportion by number and area of Type I and Type II fibers was calculated for each architecturally distinct region (superficial, intermediate and deep) and fascicle of

LMT. Percentage of Type I and Type II fibers within architecturally distinct regions and fascicle were characterized using descriptive statistics (mean, standard deviation).

Regional differences with LMT were analyzed using analysis of variance (ANOVA).

Post hoc tests were used to explore for a relationship between any of the independent variables (diameter, fiber type proportion by number, and fiber type proportion by area) for each region of LMT. Paired t-tests were carried out to compare means. Statistical significance was set at p<0.05.

Chapter 5 Results

5.1 Morphology and Architecture of Lumbar Multifidus

The unique modeling technique used in this study allowed both visualization and quantification of musculotendinous morphology and architecture. The LMT was found to consist of five bands bilaterally for all ten specimens studied. Each band originated from a lumbar vertebra (L1-L5). Muscle fiber bundles within each band having similar superior and inferior points of attachment were used to identify architecturally distinct regions. The L1-L4 bands consisted of three regions: superficial, intermediate, and deep

(Figure 5.1A) while the L5 band consisted of only two regions: superficial and deep

(Figure 5.1D).

5.1.1 Superficial LMT

The fiber bundles of superficial LMT attached superiorly via a common tendon to the tips of the spinous processes (L1- L5) and passed inferolaterally to the mammillary processes of L5, S1, the sacrum and ilium. The portion of superficial LMT originating from:

L1 spinous process had three fascicles2 attaching inferiorly to the L5 and S1

mammillary process and posterior superior iliac spine (PSIS) respectively (Fig. 5.2A).

L2 spinous process had two fascicles, attaching inferiorly to the S1 mammillary

process and PSIS respectively (Fig. 5.2B).

2 Fascicles are collections of muscle fiber bundles with the same origin and insertion (Adams, 2002).

72 73

L3 spinous process had one fascicle which attached inferiorly to the dorsolateral

aspect of the sacrum between the first to third sacral segments (Fig. 5.2C).

L4 spinous process consisted of one fascicle which attached to the posterior surface of

the sacrum between the second to fourth sacral segments, medial to the inferior

attachment of L3 fascicle (Fig. 5.2D).

L5 spinous process also consisted of one fascicle which attached to the posterior

surface of the sacrum between the level of the third to fourth sacral segments, lateral

to the median crest of the sacrum but medial to the inferior attachment of L4 fascicle

(Fig. 5.2E).

5.1.2 Intermediate LMT

Intermediate LMThad a muscular superior attachment to the spinous processes of

L1-L4. Inferiorly, L1, L2 and L3 portions attached via tendons to the L4, L5 and S1 mammillary processes respectively. However, the L4 portion attached onto the sacrum at the S2 level (Fig. 5.1B). The intermediate LMT was absent at L5 in all ten specimens

(Fig. 5.1D) being replaced by loose fatty tissue.

5.1.3 Deep LMT

Deep LMT consisted of five fascicles (L1-L5) which where entirely muscular.

Each fascicle attached superiorly to the lamina of the lumbar vertebra L1-L5, and inferiorly to the L3- S1 mammillary process two levels inferior to the superior attachment. The L5 fascicle attached to the sacrum (Fig. 5.1C). Segmental fatty replacement was observed in three specimens.

Figure 5.1. Digitization and three dimensional modeling of superficial, intermediate and deep regions of lumbar multifidus (LMT) of a cadaveric specimen, lateral views. A: Three dimensional reconstructions of the superficial (red), intermediate (yellow) and deep (purple) regions of LMT attaching to the L1 vertebra. B: Fascicles of the intermediate region attaching to the L1-L4 spinous processes. Note that there is no intermediate LMT attaching to the spinous process at L5. C: Fascicles of the deep region attaching to the L1-L5 laminae. D: Regions of LMT attaching to the L5 spinous process. Intermediate LMT is absent at L5. Spinous process (sp); mammillary process (mp); posterior superior iliac spine (PSIS); Lumbar (L).

Figure 5.2. Digitization and three dimensional modeling of superficial segments of lumbar multifidus (L1-L5), of a cadaveric specimen, lateral views. A, B, C, D, E: Fascicles of the superficial region attaching to L1-L5 spinous processes. Spinous process (sp); mammillary process (mp); posterior superior iliac spine (PSIS); Lumbar (L).

5.1.4 Architectural parameters

Mean FBL3 was significantly different between the three regions (p ≤ 0.05).

Deep LMT had the shortest mean FBL (2.9 ± 1.1cm) while superficial had the longest

(5.8 ±1.6cm). Intermediate LMT had a mean FBL of 4.1 ± 1.5cm. Comparison of FBL among levels L1 through L5 within a given region was shown to be significant for the superficial region only. Fiber bundles attaching superiorly to the L1 and L2 spinous processes had the longest FBL, while those attaching to the L4 and L5 spinous processes were the shortest (Table 5.1).

LMT Mean FBL (cm) (n=10) Fascicle(s) Level Region Superficial to L5 6.0 ±1.8 L1 to S1 6.9 ±1.7 7.3 ±1.7a to PSIS 8.4 ±2.0 L2 to S1 5.5 ±1.5 6.4 ±1.0a,b 5.8 ± 1.6a to PSIS 6.8 ±1.2 L3 to Sa * 5.6 ±1.1b,c L4 to Sa * 4.8 ± 1.2c L5 to Sa * 4.8 ± 1.7c Intermediate L1 to L4 * 3.9 ±1.7d L2 to L5 * d 4.4 ±1.8 b L3 to S1 * 3.9 ±1.6d 4.1 ± 1.5 L4 to Sa * 4.1 ± 0.9d L5 absent absent absent Deep L1 to L3 * 2.6 ±0.6e L2 to L4 * 2.7 ±0.8e 2.9 ± 1.1c L3 to L5 * 2.6 ±0.8e L4 to S1 * 3.0 ± 1.3e L5 to Sa * 3.6 ± 1.4e Table 5.1. Summary of Mean FBL for LMT. The superscripts letters are used to indicate the presence or absence of statistical significance (analysis of variance) between the three regions or among levels within a region. If the superscripts in a column differ, then the result is statistically significant. If the letter is repeated, there is no statistical significance. PSIS, posterior superior iliac spine; Sa, sacrum; FBL, mean fiber bundle length; *, only one segment present at these levels, hence mean FBL for segment equals mean FBL for level.

3 FBL = Fiber bundle length.

In contrast to mean FBL, changes in mean FBA4 were not significant (p= 0.0869), but increased from the superficial to deep regions of LMT (Table 5.2). The average FBA calculated for the superficial, intermediate and deep regions of LMT for levels L1 –L4 ranged from 11.7º to 17.2º. The largest angle was observed within the deep region at the

L5 level (28.0 ± 11.8º).

LMT Mean FBA (º) (n=10) Fascicle Level Region (s) Superficial to L5 14.3 ± 8.0 L1 to S1 12.4 ± 8.9 14.5 ±7.7a to PSIS 16.4 ± 11.8

L2 to S1 12.5 ± 3.9 a 14.4 ±5.1a 13.7 ±6.9 to PSIS 16.0 ± 6.2 L3 to Sa * 11.7 ±5.4a L4 to Sa * 12.6 ± 8.0a L5 to Sa * 15.3 ± 8.4a Intermediate L1 to L4 * 15.9 ±6.7a L2 to L5 * 14.0 ±3.9a 15.3 ±7.0a L3 to S1 * 16.4 ±8.9a L4 to Sa * 14.8 ±8.6a L5 absent absent absent Deep L1 to L3 * 13.1 ±5.0a L2 to L4 * 15.9 ±5.6a,b 18.3 a L3 to L5 * 17.2 ±12.1a,b ±10.4 L4 to S1 * 16.7 ±10.1a,b L5 to Sa * 28.0 ±11.8b

Table 5.2. Summary of Mean FBA for LMT. The superscripts letters are used to indicate the presence or absence of statistical significance (analysis of variance) between the three regions or among levels within a region. If the superscripts in a column differ, then the result is statistically significant. If the letter is repeated, there is no statistical significance. PSIS, posterior superior iliac spine; Sa, sacrum; FBA, fiber bundle angle; *, only one segment present at these levels, hence mean FBA for segment equals mean FBA for level.

Average volumes for the superficial, intermediate and deep region were significantly different from one another and decreased from superficial to deep (Table 5.3). Within

4 FBA = Fiber bundle angle.

the superficial region, the mean volume of the L1 to L3 levels was significantly larger than the L4 and L5 levels. In the deep region, the L5 level had a significantly larger volume compared with the L1 to L4 levels. No significant difference in mean volumes was noted within the levels of the intermediate region.

LMT Mean Volume (ml) (n=10) Fascicle Level Region (s) Superficial to L5 1.1 ±0.2 L1 to S1 1.6 ±0.6 6.7 ±0.5a to PSIS 3.9 ±0.7 L2 to S1 1.6 ±0.4 5.6 ±2.3a 7.9 ±1.9a to PSIS 6.3 ±1.6 L3 to Sa * 7.0 ±1.7a L4 to Sa * 3.7 ±0.4b L5 to Sa * 2.8 ±0.3 b Intermediate L1 to L4 * 1.7 ±0.4a L2 to L5 * 1.9 ±0.3a 1.7 ±0.4b L3 to S1 * 1.5 ±0.3a L4 to Sa * 1.8 ±0.5a L5 absent absent absent Deep L1 to L3 * 0.5 ± 0.1a a L2 to L4 * c 0.6 ±0.2 0.7 ±0.3 L3 to L5 * 0.6 ±0.2a L4 to S1 * 0.5 ±0.1a L5 to Sa * 1.3 ±0.1b

Table 5.3. Summary of Mean Volume for LMT. The superscripts letters are used to indicate the presence or absence of statistical significance (analysis of variance) between the three regions or among levels within a region. If the superscripts in a column differ, then the result is statistically significant. If the letter is repeated, there is no statistical significance. PSIS, posterior superior iliac spine; Sa, sacrum; *, only one segment present at these levels, hence mean volume for segment equals mean volume for level.

K-means clustering verified that FBL data could be grouped into three distinct regions

(Fig 5.3).

Figure 5.3. Histogram of fiber bundle length (FBL). K-means analyses using FBL confirmed the data could be grouped into three distinct regions.

5.1.5 Tendon architecture

The superficial and intermediate regions of LMT were both found to have extramuscular tendons associated with their bony attachments, whereas the deep region was entirely muscular. Each fascicle of superficial LMT had a common tendon which attached superiorly to the L1 to L5 spinous processes (Fig 5.2A-5.2E). In contrast to the superficial region, the tendons associated with the intermediate region were located inferiorly attaching each fascicle to the mammillary processes of L4, L5, S1 and the sacrum (Fig 5.1B). The tendons of the superficial region were thick and cylindrical, while the tendons of the intermediate region were thin and flat. Although the shape of the

80 tendinous components of the superficial and intermediate regions differed, the average tendon lengths were similar, 2.4 ± 0.1cm and 2.3 ± 0.6 cm respectively (Table 5.4).

Region Mean TL Mean FBL TL + FBL* (n=10) (cm) (cm) (cm) Superficial 2.4 ± 0.1 5.8 ± 1.6 8.2±1.6 L1 2.7 ± 0.8 7.3 ±1.7 10.0 ± 1.9 L2 2.4 ± 0.9 6.4 ±1.0 8.8± 1.4 L3 2.4 ± 0.8 5.6 ±1.1 7.9 ± 1.4 L4 2.4 ± 0.8 4.8 ± 1.2 7.2 ± 1.4 L5 2.4 ± 0.7 4.8 ± 1.7 7.2 ± 1.8 Intermediate 2.3 ± 0.6 4.1 ± 1.5 6.4 ±1.6 L1 2.0 ± 0.9 3.9 ±1.7 5.9 ± 1.9 L2 2.1 ± 0.6 4.4 ±1.8 6.5 ± 1.9 L3 2.0 ± 0.7 3.9 ±1.6 5.9 ± 1.8 L4 3.3 ±0.9 4.1 ± 0.9 7.3 ± 1.3 L5 absent absent absent Table 5.4. Tendon length, FBL and muscle lengths of superficial and intermediate regions. TL, tendon length; FBL, fiber bundle length; * , muscle length = TL + FBL.

5.2 Innervation of LMT

5.2.1 3D Model

A 3D model of the neural distribution of the medial branch to LMT was reconstructed for each specimen using the digitized data. The model could be manipulated to view the nerve distribution throughout its course from different perspectives by rotating (Figure 5.4), magnifying (Figure 5.5), and/or specifying specific regions to include/exclude (Figure 5.6). Thus, the medial branch and all of its colateral branches could be studied individually or in groups. In addition color coding of the individual branches within the different regions of LMT aided in clarifying innervation patterns.

Figure 5.4. Views of the nerve supply to lumbar multifidus (LMT) by rotation of model.

Figure 5.5. Views of the nerve supply to lumbar multifidus (LMT) at different magnifications.

Figure 5.6. Views of the nerve supply to specific regions of lumbar multifidus (LMT).

5.2.2 Nerve distribution through LMT

The innervation pattern described below for LMT is based on meticulous dissection and digitization of three cadaveric specimens. The distribution of the medial branch of posterior rami L1-L5 and all of its co-lateral branches was repeatable in all specimens investigated. Hence, the descriptions of the innervation pattern to LMT which follow are representative of all specimens used in this study.

The five bands of LMT which originated from each lumbar vertebra (L1-L5) were supplied by the medial branch of the posterior ramus of its corresponding spinal nerve.

For example, the band of LMT (which contained deep, intermediate and superficial regions) attaching onto the L1 vertebra was supplied by the medial branch of the posterior ramus of the L1 spinal nerve which exited from the L1-L2 intervertebral foramen (Figure 5.7).

From its origin to the posterior ramus, each medial branch traveled inferomedially crossing over the intersection formed between the root of the next inferior transverse process and superior articular process (Figure 5.8 & 5.9). For example the L1 medial branch was observed to cross the root of the inferior transverse process and superior articular process of L2. The medial branch continued its course inferiorly passing through a tunnel formed by the mamillary process, the accessory process and mamillo- accessory ligament (Figure 5.10). In this region the medial branch was tightly bound to the underlying bone by strong bands of connective tissue. Upon exiting this tunnel, the medial branch gave off a small collateral branch. This collateral branch traveled in an inferomedial direction before turning sharply posterior (Figure 5.10) to enter deep LMT.

This small branch was observed to enter the muscle belly on its deep surface close to its inferior attachment to the mamillary process two vertebral levels inferior (Figure 5.11).

After giving off the branch to deep LMT, the main trunk of the medial branch continued its course medially curving around the root of the superior articular process on route towards the base of the spinous process (Figure 5.12). For example, the L1 medial branch crossed over the root of the superior articular process of L2 towards the base of the L2 spinous process. From the main trunk of the medial branch, a small branch was traced superiorly to supply the zygapophyseal joint (Figure 5.12).

After giving off this articular branch, the main trunk traveled further inferomedially and divided into three separate branches (Figure 5.13 and 5.14). One branch was traced posteriorly to supply the interspinous ligament, a second continued inferiorly to enter intermediate LMT, and a third branch traveled further posteriorly and inferiorly to supply separate fascicles of superficial LMT. For example, the L1 medial branch supplied the fascicles of superficial LMT originating from the L1 spinous process and inserting inferiorly onto the L5 and S1 mamillary processes, and the sacrum (Figure 5.15 and

5.16). In a similar fashion, the L2 medial branch supplied the fascicles of superficial

LMT originating from the L2 spinous process and inserting inferiorly onto the S1 mamillary process and sacrum. Superficial LMT originating from lumbar spinous processes of L3 and L4 consisted of only one fascicle each; hence only one nerve branch per fascicle was observed. In all instances, these nerve branches traveled longitudinally along the undersurface of each fascicle. Smaller branches were observed to radiate outwards from each main stem to penetrate and supply each fascicle.

The pattern of innervation outlined above was characteristic of medial branches L1 through L4. The L5 medial branch in contrast arose from the L5 spinal nerve and from here traveled inferiorly, passing deep to the lumbosacral zygapophyseal joint. The nerve was traced posteromedially before it bifurcated to give rise to two separate branches.

One branch traveled in a posterior direction to enter deep LMT originating from the lamina of L5 vertebra (Figure 5.17). The remaining branch coursed in a posterior and slightly medial direction towards the superficial LMT originating from the spinous process of L5. This branch traveled on the deep surface of the fascicle giving off numerous small muscular branches along its course. Since intermediate LMT is absent at

L5, the medial branch which took its origin from the L5 spinal nerve was observed to innervate only the deep and superficial regions of the LMT.

Figure 5.7. Lateral view of lumbosacral spine showing medial branches (L1 to L5) which supply the five bands of LMT. L, lumbar; sp, spinous process.

86 87

Figure 5.9. Dissection of lumbar multifidus showing extramuscular course of L1 medial branch, right lateral view. The L1 medial branch is shown crossing over the intersection formed between the root of the next inferior transverse process and corresponding superior articular process (shaded blue area). Superficial (red), intermediate (blue), and deep (yellow) regions of LMT attaching to the L1 spinous process and lamina. L, lumbar; LMT, lumbar multifidus; mp, mammillary process; sp, spinous process; t, tendon; tp, transverse process.

Figure 5.10. Close up lateral view of the L2 lumbar vertebra and L1 medial branch. L1 medial branch is shown passing beneath the mamillo-accesory ligament (denoted with asterisks, “*”). Solid red line (a): branch supplying deep LMT; solid blue line (b): continuation of medial branch. L, lumbar; ap, accessory process; mp, mamillary process; sap, superior articular process; tp, transverse process.

Figure 5.11. Dissection of lumbar multifidus, right lateral view showing main trunk of L1 medial branch (yellow dotted line) giving off a nerve branch to supply deep LMT (a: red dotted line).Continuation of L1 medial branch (b: blue, dotted line).Superficial (red shaded area), intermediate (blue shaded area), and deep (yellow shaded area) regions of LMT attaching to the L1 spinous process and lamina.L, lumbar; LMT, lumbar multifidus; tp, transverse process; sap, superior articular process.

Figure 5.12. Lateral view of lumbar spine showing L1 medial branch giving off articular branch (solid red line) to supply the superior zygapophyseal joint. Solid blue line denotes continuation of medial branch. L, lumbar; iap, inferior articular process; mp, mamillary process; sap, superior articular process; tp, transverse process.

Figure 5.13. Lateral view of the lumbar spine showing medial branch dividing into three branches. Solid green line (a): branch supplying interspinous ligament; solid red line (b): branch supplying superficial LMT; solid blue line (c): branch supplying intermediate LMT. L, lumbar; iap, inferior articular process; mp, mamillary process; tp, transverse process; sap, superior articular process.

Figure 5.14. Dissection of lumbar multifidus (LMT), right lateral view showing medial branch (mb) of posterior ramus giving off branches to supply superficial (red), and intermediate (blue) regions of LMT attaching to the L1 spinous process and laminae. Deep LMT has been removed. Ib, branch to intermediate LMT (solid blue line); sb, branch to superficial LMT (dashed red line); branch to interspinous ligament (dashed green line); ab, branch to L1-2 zygapophyseal joint (solid pink line); L, lumbar; sp, spinous process; tp, transverse process; ZJ, zygapophyseal joint.

Figure 5.15. Lateral view of digitized spine showing the L1 medial branch dividing into three branches to supply the three separate fascicle of superficial LMT attaching superiorly to the L1 spinous process. L, lumbar; sp, spinous process; tp, transverse process; iap, inferior articular process; sup, superior articular process.

Figure 5.16. Dissection of lumbar multifidus (LMT), right lateral view showing medial branch (mb) of posterior ramus giving off branches to supply fascicle of superficial LMT attaching superiorly to the L1 spinous process. Deep LMT has been removed. Intermediate LMT is reflected. sb, branch to superficial LMT; L, lumbar; mb, medial branch; sp, spinous process; tp, transverse process; ZJ, zygapophyseal joint; iap, inferior articular process; sup, superior articular process.

Figure 5.17. Lateral view of lumbosacral spine showing the L5 medial branch innervating deep and superficial regions of LMT. L, lumbar; iap, inferior articular process; sap, superior articular process.

5.3 Mean characteristics of muscle fiber type for LMT: Pilot Study

The results for fiber type distribution and fiber type diameter are based on 29 muscle biopsies taken form one fresh male cadaveric specimen and therefore cannot be generalized to the population.

5.3.1 Fiber type distribution

For the muscle as a whole, the average number Type I muscle fibers in LMT expressed as a percentage was 60%. Similarly, the average area occupied by Type I muscle fibers expressed as a percentage was 70% (Table 5.5).

LMT Number of Type I fibers expressed Area of Type I fibers expressed as a proportion as a proportion (n=1) Spinal Fascicle(s) Spinal Level Region Total Fascicle(s) Region Total Level Superficial to L5 0.71 ± 0.13 0.73 ± 0.09 L1 to S1 0.61 ± 0.61 0.58 ± 0.14 0.66 ± 0.06 0.64 ±0.10 to PSIS 0.43 ± 0.02 0.62 0.54 ± 0.03 0.75 L2 to S1 0.69 ± 0.13 ± 0.80 ± 0.001 ± 0.68 ± 0.09 0.81 ± 0.01 to PSIS 0.67 ± 0.06 0.14 0.81 ± 0.006 0.11 L3 to Sa * 0.73 ± 0.06 * 0.86 ± 0.02 L4 to Sa * 0.65 ± 0.28 * 0.82 ± 0.13 L5 to Sa * 0.49 ± 0.10 * 0.74 ± 0.04 Intermediate 0.60 0.70 L1 to L4 * 0.66 ± 0.06 * 0.75 ± 0.04 0.63 ± 0.70 ± L2 to L5 * 0.61 ± 0.19 * 0.68 ± 0.11 ± 0.12 ± 0.12 L3 to S1 * 0.58 ± 0.04 * 0.69 ± 0.05 0.09 0.06 L4 to Sa * 0.69 ± 0.02 * 0.67 ± 0.02 L5 - - - - - Deep L1 to L3 * 0.52† * 0.52† 0.49 0.56 L2 to L4 * 0.47† * 0.53† † ± † ± L3 to L5 * 0.51 0.03 * 0.54 0.07 L4 to S1 * 0.45† * 0.54† L5 to Sa * 0.48† * 0.69† Table 5.5. LMT Type I fiber proportions (mean ± SD). † no standard deviation as only one biopsy sample taken from the deep region of LMT. PSIS, posterior superior iliac spine; Sa, sacrum; FBA, mean fiber bundle angle; X, absent at this level; *, only one segment present at these levels, hence mean FBA for segment equals mean FBA for level.

The area occupied by Type I fibers was significantly different between the deep, intermediate and superficial regions of LMT (p = 0.038). Figure 5.18 shows the relationship between the proportion of Type I fiber area and region for LMT. A Tukey’s post hoc test showed that the area occupied by Type I fibers was significantly different between the deep and superficial regions (p = 0.012) but not between the deep and intermediate levels (p = 0.09) or superficial and intermediate levels (p = 0.47) (Figure

5.19).

Figure 5.18. Mean area of Type I fibers for each region expressed as a proportion. If the letter(s) over the error bars are duplicated, then there is no statistical difference. If the letter(s) over the error bars are different, then there is a statistical difference.

Figure 5.19. Comparison of the mean area occupied by Type I fibers between the deep, intermediate and superficial regions of LMT. A. The area occupied by Type I fibers (brown) was significantly different between the deep and superficial regions (p< 0.01); B-C. The area occupied by Type I fibers was not significant between the deep and intermediate and intermediate and superficial regions (p> 0.05).

In contrast to the area occupied by Type I fibers, the number of Type I fibers present within the LMT was not affected by level (L1-L5), region (deep, intermediate or superficial), or biopsy location (superior, middle or deep).

The area occupied by Type I fibers for individual fascicle of LMT varied significantly across different fascicles of LMT (p = 0.019). Figure 5.20 shows the relationship between the proportion of Type I fibers and the various fascicles of LMT.

Figure 5.20. Mean areas of Type I fibers expressed as a proportion. Notice that within each band of LMT the area occupied Type I fibers increases from deep to superficial. The dotted line joining points helps illustrate this interesting “sawtooth” pattern and is not intended to depict continuous data.

The area of Type I fibers is observed to increase from the deep to superficial regions of LMT within a given spinal level (e.g. L1), with this upward trend repeating itself from L1-L5. When the fascicles of LMT are arranged from superior to inferior along the horizontal axis of the scatter plot above, an interesting sawtooth pattern emerges. Hence, the LMT is comprised of five primary bands (L1–L5) based on the area occupied by Type I fibers within each fascicle of LMT. Each of these five bands is supplied by its own medial branch of the posterior ramus as described previously in section 5.2.2.

5.3.2 Fiber type diameter

The average diameter of Type I muscle fibers for deep, intermediate and superficial LMT were 59.6µm, 56.8µm, and 55.5µm respectively (differences between regions shown to be insignificant, p = 0.94). The corresponding values of Type II fibers were 50.6µm, 49.7µm and 41.5µm, respectively, and were shown to be significantly different from each other (p< 0.005) (Table 5.6).

LMT Type I fiber diameter (um) Type II fiber diameter (um) (n=1) Segment(s) Level Region Segment(s) Level Region Superficial to L5 48.95 ± 8.43 52.77 ± 8.59 47.18 ± 7.51 47.01 ± 8.32 L1 to S1 56.15 ± 8.09 49.49 ± 9.86 to PSIS 53.20 ± 8.06 44.35 ± 6.92 55.51a 41.47a L2 to S1 53.01 ± 15.98 53.07 ± 13.72 ± 10.92 40.64 ± 9.62 38.80 ± 10.63 ± 10.45 to PSIS 53.13 ± 14.43 36.94 ± 11.51 L3 to Sa * 61.07 ± 10.62 * 41.26 ± 9.85 L4 to Sa * 60.81 ± 8.27 * 37.14 ± 8.78 L5 to Sa * 57.77 ± 9.86 * 33.93 ± 9.92 Intermediate L1 to L4 * 57.81 ± 10.94 * 46.59 ± 8.52 L2 to L5 * 57.53 ± 8.61 56.82a * 50.24 ± 9.38 49.69b L3 to S1 * 56.71 ± 8.45 ± 8.97 * 44.03 ± 10.37 ± 10.77 L4 to Sa * 55.24 ± 8.06 * 57.89 ± 9.95 L5 - * * - Deep L1 to L3 * 53.59 ± 7.76 * 53.80 ± 8.59 L2 to L4 * 46.71 ± 10.97 59.63a * 40.83 ± 8.41 50.56c L3 to L5 * 57.03 ± 11.05 ± 14.88 * 54.40 ± 7.99 ± 10.23 L4 to S1 * 64.37 ± 9.77 * 54.48 ± 12.27 L5 to Sa * 76.46 ± 14.98 * 49.27 ± 7.66 Table 5.6. LMT fiber type diameters (mean ± SD). PSIS, posterior superior iliac spine; Sa, sacrum; FBA, mean fiber bundle angle; X, absent at this level; *, only one segment present at these levels, hence mean FBA for segment equals mean FBA for level. Type II fiber diameters were significantly different across different regions of LMT, p < 0.005.

Differences in Type I and Type II mean fiber diameter were investigated across different regions (deep, intermediate, or superficial) and levels (L1, …, L5) of LMT, as well as biopsy location (superior, middle or inferior) for a given fascicle of LMT.

Type I fiber diameters increased as spinal level increases from L1 to L5 (Figure

5.21). Tukey’s post hoc test verified that fiber diameters measured for LMT at one vertebral level were statistically significant only if compared with fiber diameters taken from LMT two or more levels inferior. For example, the difference between the mean diameters at L1 and L2 was not statistically significant, nor was the difference between the mean diameters at L3 and L4. However, the difference in mean diameters between

L1 and L4, L1 and L5, L2 and L4, and L2 and L5 spinal levels were statistically significant.

Figure 5.21. The relationship between mean Type I cell diameters and spinal level. If the letter(s) over the error bars are duplicated, then there is no statistical difference. If the letter(s) over the error bars are different, then there is a statistical difference.

The size of Type I fibers within a given region of LMT depended on spinal level

(p = 0.013). This relationship is represented graphically Figure 5.22 and indicates that deep LMT at L5 had the largest diameters.

Figure 5.22. The relationship between mean Type I fiber diameters and spinal level and region.

Type II fiber diameter decreased from the deep to superficial regions of LMT

(p<.0005). This relationship is depicted graphically in Figure 5.23. Tukey’s post hoc test showed that the mean diameter of Type II fibers differed significantly between the deep and superficial regions, and the intermediate and superficial regions of LMT (p < 0.001).

Biopsy location (i.e. superior, middle, inferior) did not significantly affect the fiber diameters of Type II fibers (p = 0.26).

Figure 5.23. The relationship between mean Type II fiber diameters and region. If the letter over the error bars is the same, then there is no statistical difference. If the letter over the error bars is different, then there is a statistical difference.

Chapter 6 Discussion

6.1 Introduction

There has been a paucity of studies in the literature that concurrently examine the morphology, architecture, fiber type distribution and nerve innervation pattern of the muscles of the lower back throughout their entire volume. In order to begin the work of examining the macro and micro anatomy of the back muscles in this manner, the largest and most medial back muscle in the lumbosacral region, lumbar multifidus (LMT) was selected for this study.

The importance of studying how LMT assists in stabilizing the lumbar spine cannot be understated. The pioneering works of Crisco and Panjabi (Panjabi et al., 1989; Crisco

& Panjabi, 1991) have demonstrated that deep segmental muscles, such as LMT, can significantly reduce intersegmental movement and hence augment stability. Clinically, it has been shown that following an acute episode of LBP, one is 12.4x more likely to suffer a reoccurrence if LMT is inadequately rehabilitated (Hides et al., 2001). In addition, surgery performed on individuals with LBP may damage LMT muscle architecture as well as its nerve supply potentially leading to functional deficits (Ward et al., 2009), including lumbar instability. Failed back surgery may ironically be due to the iatrogenic effect of disrupting the morphology, architecture, and/or nerve supply to LMT as a consequence of surgery itself.

The morphology, nerve supply and fiber type characteristics of LMT have been previously reported (Bogduk et al., 1982; Ford et al., 1983; Macintosh et al., 1986;

Rantanen et al., 1994; Bajek et al., 2000; Zhao et al., 2000; Yoshihara et al., 2001) but not

100 101 through its entire volume. The current study, through meticulous dissection and digitization of individual muscle fiber bundles and the medial branches to each of the five bands of LMT has contributed to our understanding of LMT function. In addition, the immunohistochemistry portion of this study which examined the distribution and size of

Type I and II muscle fibers throughout the volume of LMT has yielded some interesting results which support the theory that LMT is subdivided into neuromuscular compartments.

6.2 Morphology and Architecture

The morphology and architecture of LMT were captured throughout its volume by digitizing up to 1400 individual muscle fiber bundles per specimen. This method, unlike previous studies which simplified the morphology and architecture of LMT using straight lines to represent muscle fascicles and their line(s) of action (Macintosh & Bogduk, 1986;

Macintosh et al., 1986) allows individual muscle fiber bundles to be three dimensionally visualized and their architectural characteristics (i.e. fiber bundle length and fiber bundle angle) to be quantified. The three dimensional models of the LMT produced in this manner enables viewing and quantification of muscle architecture at a level of complexity which could not be achieved previously (Agur et al., 2003; Kim, Boynton et al., 2007)

As highlighted earlier, muscle morphology and in particular, architectural parameters such as FBA, FBL, and muscle volume are important determinants of function which can have significant effects on muscles’ force generating capability (Roy & Ishihara, 1997;

Lieber & Friden, 2000). For example, the orientation (i.e. angle) of muscle fiber bundles

102 within a muscle may be used to determine its actions (Macintosh & Bogduk, 1986) and is integral to ensuring proper electrode placement when carrying out electromyography power spectrum studies (De Foa et al., 1989; Biedermann et al., 1991).

6.2.1 Morphology

Few studies have documented the morphology of LMT with “anatomical texts and atlases describing muscle structure mainly in terms of their origins and insertions”

(Biedermann et al., 1991). In one cadaveric study (n=12), Macintosh et al (1986) found

LMT to consist of five bands (L1-L5), with each band originating from a lumbar spinous process and inserting inferiorly onto mammillary process(es), the sacrum and/or the ilium.

Similar to Macintosh et al (1986), the current study found LMT to consist of five bands; each band attaching to a separate lumbar vertebra. However, closer inspection revealed hierarchical arrangement of muscle fiber bundles. Muscle fiber bundles were organized into groups to form fascicles with specific superior and inferior points of attachment onto the lumbosacral spine. Fascicles in turn originating from L1 to L4 vertebrae were further organized into distinct regions: deep, intermediate or superficial.

Whereas the fascicles attaching to the L5 vertebra were organized into only two regions: superficial and deep. The rationale for the absence of an intermediate region at L5 is not entirely clear, but may have some influence on the higher incidence of pathology (e.g. herniated discs) and instability (e.g. degenerative spondylolisthesis) observed at least at the lumbosacral junction (Boden et al., 1996).More studies are needed to determine if intermediate LMT is absent at L5 across different age cohorts.

103

Furthermore, only the intermediate and superficial regions of LMT contained tendons within their structure. The placement and gross morphology of tendons within a muscle influences its function (Lundon, 2003). Hence, the position of tendons relative to contractile tissues may influence the role that these two regions play in helping to stabilize the lumbar spine. The intermediate regions of LMT contained tendons that were both broad and flat and which seemed to anchor muscle fibers to bone inferiorly. In direct contrast, the tendons associated with the superficial regions of LMT tended to be thick and cylindrical, attaching muscle fiber bundles of LMT superiorly to the tip of the spinous processes L1-L5. This arrangement of tendons within the structure of LMT may have the following functional interpretation. The intermediate regions of LMT - with its superiorly placed muscle fiber bundles and inferiorly placed tendons - have a line of action that follows a caudal to cranial direction. This arrangement seems appropriate in the intersegmental control of anterior translation. The superficial regions of LMT - with its inferiorly placed muscle fiber bundles and superiorly positioned tendons – have a line of action that is in a cranial to caudal direction. This arrangement is better suited in helping maintain the lumbar lordosis, particularly with forward bending (Figure 6.1).

104

Figure 6.1. Lateral view of lumbosacral spine showing superficial and intermediate regions of LMT attaching to the L1 spinous process. Yellow arrows depict the direction of the line of action for each region.

6.2.2 Measurement of architectural parameters of LMT

The current study quantified key architectural parameters for LMT according to its fascicular and regional arrangement of muscle fiber bundles. Differences in FBL, FBA and muscle volume have important functional implications which are discussed in the sections which follow. Before proceeding with this discussion, it is prudent to point out some potential factors that could affect measurements taken from cadaveric specimens.

Cutts (1988) examined the effects of fixation on skeletal muscle tissue and found that a significant loss in muscle length occurs when muscles are removed and measures before fixation compared to those measured in situ while still attached to the bony skeleton. The current study measured the FBL and FBA of individual muscle fiber bundles using the digitizer directly (i.e. in situ) for each specimen. Although this reduced the possibility of whole muscle length shrinkage and hence FBL/FBA changes due to fixation, it is still possible that the embalming procedure could have altered these intrinsic architectural

105 properties. In addition, “long-term degradation of muscle tissue may slowly, throughout the course of storage, alter the architecture within skeletal muscle”(Martin et al., 2001).

6.2.2.1 Fiber Bundle Length (FBL)

Delp et al (2001) provided quantitative descriptions of muscle architecture for the rectus abdominis, quadratus lumborum, and erector spinae muscles. Musculotendon lengths, muscle lengths, fascicle lengths, sarcomere lengths, pennation angles, and muscle masses were measured in five cadavers. While Delp et al (2001) provided new architectural data for some of the abdominal and back muscles, these authors did not provide descriptive data on other muscles important in spinal stability such as LMT.

Bogduk et al (1992) constructed a model of the back muscles which included 49 fascicles of the lumbar erector spinae and LMT. To determine the physiological cross sectional area of individual fascicles required quantification of both their volumes and lengths. These authors explained these measurements were taken during the course of previous morphological studies of these muscles (Macintosh et al., 1986; Macintosh &

Bogduk, 1991), but acknowledge that the data had not been previously reported. In the model presented by Bogduk et al (1992) LMT consisted of 11 separate fascicles with fiber bundle lengths ranging from 41 mm to 190 mm. The average FBL for each fascicle of LMT in the current study was smaller than reported by Bogduk et al (1992). This is likely due to the fact the values reported for FBL in the current study are of only the contractile portions of each fascicle and does not include the length of any of its tendons.

Ward et al (2006; 2009) and Kim, Gottschalk et al (2007) studied the architectural features of LMT in human cadavedic spines from T12 to the sacrum. LMT muscles were isolated and measurements taken of muscle mass, ML, and FBL. “To compensate for

106 variations in raw fiber length that occur because of the position of the spine during fixation, muscle fiber lengths were normalized by scaling to the optimal sarcomere length of human muscle (2.7µm)”(Ward et al., 2009). These authors reported the average FBL of LMT as a whole to be approximately 5.66 ±0.65cm. In addition, these authors found significant difference in FBL across segmental levels of origin with a trend towards increasing FBL from T12 to L2 and then decreasing to L5.

In the present study, LMT shows a hierarchical arrangement of muscle fiber bundles. As described previously, individual fiber bundles were arranged to form fascicles, which in turn were organized into one of three distinct regions: superficial, intermediate or deep. Average FBL measured ranged from 2.6 ±0.6 cm to 7.3 ±1.7 cm

(see Table 5.1) with superficial LMT demonstrating the largest average FBL (5.8 ±

1.6cm). The standard deviation calculated for each region and fascicle of LMT was relatively small suggesting little variation over the ten specimens studied for this architectural parameter.

The average FBL of superficial LMT determined in this study appear to corroborate the earlier work of Ward et al (2009) for the LMT muscle as a whole. In contrast however, the current study showed that FBL decreased from L1 to L5 for the superficial region opposed to the increasing then decreasing trend reported by Ward et al

(2009) (Figure 6.3). Fascicles containing fibers with large FBL are designed to produce more excursion (Ward et al., 2009). Hence, the decrease in FBL demonstrated from L1 to L5 in this study may reflect the greater importance of fiber bundles/fascicles originating from more superior vertebrae (e.g. L1 and L2) in helping to maintain and control the lumbar lordosis. The corollary of this is that fiber bundles/fascicles

107 originating from inferior lumbar vertebrae (e.g. L4, L5) produce a smaller excursion and hence are more important in controlling intersegmental movement.

(FBL) 4 L1 L2 L3 L4 L5 L1 L2 L3 L4 L5 2 Fiber Bundle Length Length Bundle Fiber 0 Ward et al, 2009 Rosatelli et al, 2008 Spinal Level

Figure 6.2. Bar graph showing comparison of FBL values as a function of segmental level.

6.2.2.2 Fiber Bundle Angle (FBA)

Using photographs taken in the plane of the back of both male and female cadaveric specimens DeFoa et al (1989) and Biedermann et al (1991) determined the fiber direction of longissimus, iliocostalis and lumbar multifidus by measuring the angle subtended between anatomical reference lines and the spine. For LMT, the FBAs’ reported for male and female cadaveric specimens was 15.1° (range: 13.5°-18.0°) and 23.5° (17.5°-28.5°) respectively (see Table 2.2). Although these authors did not specify from what region of

LMT (e.g. deep vs superficial) measurements were taken, one may speculate that the values reported are representative of the most superficial portion of LMT as these are more readily attainable. Assuming we are justified in comparing the average FBA for

108 the superficial region of LMT in the current study (13.7° ± 6.9°) with the average FBA calculated from previous studies (average male specimens: 15.1°; average female specimens: 23.5°) we notice that for male cadaveric specimens the values are comparable. A similar comparison cannot be made for the FBA of female specimens due to insufficient numbers (i.e. one female specimen in this study). Hence, it is also not possible to comment on gender differences in FBA.

By plotting the attachments of each fascicle of LMT from both cadavers (n=5) and living subjects (n=21) onto clinical radiographs, Macintosh and Bogduk

(1986)determined that the “principal action of multifidus is posterior sagittal rotation”.

The current study corroborates this function; interestingly however the current study found that the FBA for superficial LMT calculated at L5 was 10º greater than that reported by these authors (Tables 6.1). Therelatively large FBA at L5 suggests that LMT at this level is also important in controlling rotational movement in the transverse plane at the lumbosacral junction.

As described above, Macintosh & Bogduk (1986) determined the action of LMT from the orientation of its component fascicles (Macintosh & Bogduk, 1986). Each fascicle of

LMT however consists of many hundreds of muscle fiber bundles each with their own unique FBA. Furthermore, the point of insertion of these fiber bundles may not be uniformly distributed. This suggests that the net action of a given fascicle of LMT is proportional to the spatial distribution of all its constituent muscle fibers and their associated fiber angles (Figure 6.2). The strength of this study lies in the fact that the results for mean FBA take into consideration both the number and relative distribution of

109 muscle fiber bundles throughout the entire volume of LMT. This also helps to explain the relatively large standard deviations observed for this architectural parameter.

Level FBA Ant-Post Angle Muscle Fiber Muscle Fiber (°) (°) Angulation Angulation (°) Rosatelli et al, 2008 Macintosh et al, (°) Biedermann et al, 1986 De Foa et al, 1989 1991 L1-L4 15.9 ±6.7 14.8 ± 0.8 L1-L5 14.3 ± 8.0 15.0 ± 0.7 L1-S1 12.4 ± 8.9 12.6 ± 0.6 L1-Sacrum 16.4 ± 11.8 16.6 ± 0.9 L2-L5 14.0 ±3.9 18.8 ± 1.1 Male Female L2-S1 12.5 ± 3.9 18.0 ± 1.0 15.1± 1.43 23.5 ± 4.5 (13.5-18.0) (17.5-28.5) L2-Sacrum 16.0 ± 6.2 20.0 ± 1.6 L3-S1 16.4 ±8.9 23.2 ± 1.1 L3-Sacrum 11.7 ±5.4 19.6 ± 0.9 L4-Sacrum 12.6 ± 8.0/14.8 ±8.6 15.6 ± 0.9 L5-Sacrum 15.3 ± 8.4 5.4 ± 1.5 Table 6.1. Comparison of LMT muscle fiber angles of different studies including the current.

Figure 6.3. The net action of a given fascicle of LMT is dependant on the number and distribution of muscle fibers bundles. A. Uniform number and distribution of muscle fibers (red and blue arrows). Resultant force vector (yellow arrow) assumes a median position with respect to all the fiber bundles. B. Non-uniform distribution of muscle fiber bundles (red and blue arrows). Resultant force vector is “pulled” towards the more densely packed, red arrows.

110

6.2.2.3 Volume

As described in section 6.2.2.2, Bogduk et al (1992) measured the volume of individual fascicle of LMT in order to derive their PCSA. As the authors explain, “these measurements were taken during the course of previous morphological studies of the muscle, but the data have not been reported previously”.

No other studies reported the volume of LMT as a whole or of its individual fascicles or bands. The present study fills this void. In this study, the average muscle volume for the three regions were: 5.6 ± 2.3ml for superficial LMT, 1.7 ± 0.4ml for intermediate LMT and 0.7 ± 0.3ml for deep LMT. Small amounts of fluid and connective tissue elements associated with cadaveric muscle tissue which could not be eliminated are likely to have influenced volume measurements and is considered a source of experimental error in this study.

Based strictly on volumetric data, superficial LMT because of its larger volume compared with deep and intermediate regions is better suited to generate torque and hence produce movement in the lumbar spine. Using the same train of thought, deep

LMT is least likely to produce any meaningful contraction due to its much smaller volume. In addition to having the smallest volume, deep LMT also had the shortest

FBL (see section 6.2.2.2). Based on its architecture, small FBL and volume, deep LMT is strategically positioned to provide proprioceptive feedback from the lumbar spine. The proprioceptive information that deep LMT provides may assist the central nervous system

(CNS) in regulating the correct amplitude, direction and force output of the various lumbar muscles, including of course LMT itself. Deep LMT is less likely to restrain

111 inter-segmental movement. The exception is at L5 were the volume and length of its fiber bundles is relatively greater.

6.2.2.4 Physiological Cross Sectional Area (PCSA)

At the present time, the only method of determining the force exerted by individual fascicles of the back muscles is to derive it from architectural features. In principle, the maximum force exerted by a muscle is proportional to its size: either its cross sectional area or in the case of an irregularly shaped muscle, its’ PCSA (Powell et al., 1984).

Hence, muscles with larger PCSA have greater force generating capabilities than those with smaller PCSA. In addition, it is postulated that muscles with large PCSA but small

FBL are ideally suited for providing stability (Ward et al., 2009).

By knowing a muscles volume and mean fiber length, the PCSA of a muscle can be estimated using the equation:

Muscle Volume PCSA = (Lieber & Boakes, 1988) Fiber Length

As described previously, LMT consists of muscle fiber bundles which are organized into discrete bands or fascicles. Each fascicle has specific points of attachment onto the lumbosacral spine. Hence, it is possible to represent the PCSA for a given fascicle of LMT using the equation 1.2 below:

Volume fascicle PCSA fascicle = FBL fascicle

112

In the current study the volume of each fascicle of LMT was determined using water displacement. The FBL for each fascicle was determined by averaging the lengths of individual fiber bundles. The PCSA for each fascicle of LMT was calculated and the results are compared with previously reported values in Table 6.2.

Based on PCSA, the current study revealed some interesting features about the fascicles and regions comprising LMT. It emerges that the superficial region of LMT, whose fascicles, when summed together, exhibits the greatest PCSA and hence largest total force in the sagittal plane compared with the intermediate and deep regions.

Furthermore, within the superficial region, the bands attaching to the L2 and L3 spinous processes demonstrate the largest PCSA and hence produce the greatest amount of force.

This finding may reflect the potentiality of these bands in maintaining the lumbar lordosis which is usually centered about the L3-L4 disc.

The findings of the current study are in line with those reported by Ward et al

(2009). Compared with the other back muscles the relationship between PCSA and FBL differ significantly from the other back muscles. The large PCSA and short FBL associated with LMT indicate that LMT can produce a large amount of force over a relatively small operating range. This architectural design is best suited for muscles that provide inherent stability rather than motion. Thus, the role of LMT may be to limit excessive motion across individual motion segments (disk and facets) and thereby balance the loads across the spine. This may be important to prevent adjacent level degeneration. This coincides well with previous studies that support the importance of this muscle for clinical function. Efforts to preserve multifidus muscle function, such as with minimally invasive surgical techniques, are warranted.

113

Previous investigations have shown that LMT can stiffen the spine and decrease neutral zone motion in flexion and extension (Steffen et al., 1994) and lateral flexion

(Wilke et al, 1995) but not significantly in rotation (Wilke et al, 1995). The explanation given for this finding was that LMT is at a mechanical disadvantage to control neutral zone motion in this plane. The current study found that the FBA for superficial LMT originating from L5 to be significantly larger then previously reported (Macintosh &

Bogduk, 1986). In addition the FBL measured at this level was paradoxically the shortest. Based on these findings, the current study postulates that LMT may play a larger role in controlling neutral zone motion in rotation at L5 than previously predicted by Wilke et al (1995) which only investigated the influence of muscle forces at the L4-L5 segment.

Mean Mean Mean PCSA (mm2) Mean PCSA (mm2) LMT Volume FBL Current Study Bogduk et al, 1992 (mm3) (mm) Superficial to L5 1140 ±0.2 60.0 ±18 19.0 ±2.8 42 L1 to S1 1560 ±0.6 68.8 ±17 22.6 ±9.2 36 to PSIS 3900 ±0.7 84.0 ±20 46.4 ±8.9 60 to S1 1640 ±0.4 54.6 ±15 29.8 ±7.0 39 L2 to PSIS 6300±1.6 67.6 ±12 92.6 ±24.1 99 L3 to Sa 7040 ±1.7 56.7 ±11 125.7 ±29.6 157 L4 to Sa 3660±0.4 48.1 ± 12 76.2 ±7.5 186 L5 to Sa 2820 ±0.3 42.8 ± 17 58.8 ±6.0 90 Intermediate L1 to L4 1740 ±0.4 38.6 ±17 44.6 ±3.8 40 L2 to L5 1880±0.3 43.7 ±18 42.7 ±3.8 39 L3 to S1 1460 ±0.3 39.2 ±16 37.4 ±3.8 54 L4 to Sa 1760 ±0.5 40.8 ± 9 42.9 ±3.8 186 L5 - x x x 90 Deep L1 to L3 500 ± 0.1 25.5 ±6 19.2 ±3.8 NA L2 to L4 620 ±0.2 27.4 ±8 23.0 ±5.5 NA L3 to L5 560 ±0.2 25.7 ±8 21.5 ±7.0 NA L4 to S1 500 ±0.1 30.0± 13 16.7 ±3.3 NA L5 to Sa 1260 ±0.1 36.2 ± 14 35.0 ±2.5 NA Table 6.2. Comparison of PCSA of the current and previous studies.

114

6.3 Innervation

Muscle architecture is a primary determinate of function. Hence, the more complex a muscles’ architecture, the more likely for it is to be subdivided functionally into specific regions, with each region subserving a specific role (English et al., 1993). Not surprisingly, LMT with its complex arrangement of muscle fiber bundles is ideally suited to manage multiple roles simultaneously, orchestrated by the central nervous system via an elaborate yet simplistic arrangement of nerve fibers.

As described previously, LMT is divided into architecturally distinct regions defined by fiber bundle orientation i.e. fiber bundle length, fiber bundle angle and tendinous attachments (Rosatelli et al., 2008). The muscle has three architecturally distinct regions: superficial, intermediate and deep depending on spinal level. The L1-L4 lumbar segments consisted of all three regions, while the L5 segment consisted of only two regions: deep and superficial. This unique arrangement of fiber bundles supports the hypothesis that LMT is composed of “neuromuscular compartments” which have functional or task-oriented roles. Neuromuscular compartments are defined as architecturally distinct regions within the muscle which are independently innervated by an individual nerve branch. Each compartment contains motor unit territories with a unique array of physiological attributes (English & Letbetter, 1982a). In other words, the intensity and timing of motor unit activation of each compartment can be independently controlled and can vary between regions.

The current study found that the medial branch of the posterior ramus innervated each region of LMT separately. If a region contained more than one fascicle (e.g. superficial LMT originating from the L1 and L2 spinous processes), then the medial

115 branch supplying this region would further subdivide to innervate these fascicles. Since intermediate LMT is absent at L5 it is not surprising that the medial branch supplied only its deep and superficial regions. The absence of intermediate LMT and its nerve supply at L5 may be due to changing functional demands placed on the spine as a factor of age or possibly the results of denervation brought about because of pathology.

Since it is has been established “that individual alpha motoneurons innervate muscle fibers of only single histochemical types” (English et al., 1993), this implies the muscle fiber bundles may be organized into functional units or regions. The results of this study support the hypothesis that LMT is divided into neuromuscular compartments. Each fascicle or region is controlled independently by a separate branch from the medial division of the posterior ramus, thus each region of LMT has the potential to carry out a specific function. The exact nature of this depends in turn on the arrangement of its muscle fibers (i.e. muscle architecture) and fiber type composition. Regions of LMT having large PCSA, small FBL and a relatively high concentration of Type I fibers would be expected to have a greater tonic stabilizing role.

6.4 FiberType Characteristics

Before the results of this section are interpreted it is important to address any potential methodological aspects which could affect the data. First and foremost, since only one male embalmed cadaveric specimen was used for this part of the study, the results may not be reflective of the population. In addition, the number of biopsy samples taken from each region (deep, intermediate and superficial) and vertebral level (L1-L5) was not the same. The fewest number of biopsies was taken from deep LMT, while the

116 greatest number of biopsies was taken from superficial LMT. Deep LMT being small in size precluded the taking of multiple biopsy samples of this region. In contrast, superficial and intermediate LMT constituted the bulk of LMT and therefore in order to verify homogeneity of fiber types along the length of these fascicles more biopsies were taken. Future studies making use of larger sample sizes would improve the generalizability of the results. Other factors pertinent in this study, which may have influenced the results, include: age, gender, medical history and the effect of chemical fixation on measures of cell diameter and volume/area distribution.

Several studies have been performed to investigate the relationship between age and the size, number and proportion of muscle fiber types. The majority of these studies have looked at the quadriceps muscle, particularly vastus lateralis (Porter et al., 1995).

The general conclusion is that Type II fiber size decreases with increasing age, whereas the size of Type I fibers does not appear to be affected (Porter et al., 1995). Other limb muscles have been investigated to a much lesser degree with no comparable studies performed for the back muscles. Alterations in the proportion of Type I and Type II fibers have also been reported with increasing age. Pierobon-Bormioli et al (1981) showed a higher percentage of Type I fibers taken from vastus lateralis in older subjects.

Larsson et al (1978) showed a similar result for quadriceps, but also found that the distribution of Type II fibers to decrease linearly from the third to seventh decades. Later studies seemingly contradict these earlier findings reporting no fiber type alteration with increasing age (Grimby & Saltin, 1983; Sato et al., 1984; Lexell et al., 1986). Rat soleus muscle shows an age-related decrease in the relative proportion of Type II fibers (Brown,

1987). This observed reduction may be indicative of a transformation from Type II to

117

Type I fibers secondary to chronic increased use (Lieber, 2002). Alternatively, this may be due to selective loss of Type II muscle fibers. In a series of elaborate and detailed studies on the entire vastus lateralis muscle in humans, Lexell and colleagues confirmed the large decrease in the number and total area occupied of Type II fibers (Lexell,

Henriksson-Larsen, Winblad et al., 1983; Lexell & Downham, 1992).

The specimen used in the current study was a fresh, 68 year old male cadaver.

As discussed above from previous studies, the size (i.e. diameter) of Type II fibers seems to decrease with age while the proportion of Type I fibers seems to increase. In the current study, the mean diameters of Type I fibers and Type II fibers was determined to be 56.6µm and 45.4µm respectively which is in keeping with previously recorded values

(Mattila et al., 1986; Rantanen et al., 1993; Rantanen et al., 1994; Mannion et al., 2000).

The proportion of Type I fibers was 0.60 while the proportion of Type I fiber area was

0.70. These proportions are also similar to previously reported values for the erector spinae and LMT (Mattila et al., 1986; Rantanen et al., 1993; Mannion, 1999).

Gender has also been shown to have an effect on the relative size and type distribution of the erector spinae muscles in healthy subjects and those with chronic back pain (Mannion, Dumas et al., 1997; Mannion et al., 2000). In healthy subjects, the authors provided data showing that males had larger fibers than females for each fiber type. In terms of fiber type distribution, no difference between males and females was observed for the percentage of a given fiber type, however, the percentage of Type I fiber area was significantly higher in females. Lastly, in females only, Type I fibers were found to be considerably larger compared with Type II fiber (Mannion, Dumas et al.,

1997). In the current study, a significant difference was found between the diameters of

118

Type I and II fibers (p < 0.005); however, no inference can be made as to whether gender had any influence on this result.

Injury or pathology also seems to have an effect on fiber type distribution. In patients undergoing back surgery due to herniated intervertebral discs, biopsy samples taken from LMT, demonstrate selective atrophy of Type II muscle fibers (Mattila et al.,

1986; Rantanen et al., 1993) and a significantly lower proportion of Type I fibers in individuals with low back pain (Mannion, Weber et al., 1997). However, examination of muscle biopsies taken from cadaveric specimens with no history of lumbar pathology or back pain has yielded similar results (Mattila et al., 1986; Rantanen et al., 1993).

Based on lack of relevant premortem data for the current specimen it is not possible to determine if this or the type of tissue used (i.e. cadaveric) could have influenced the studies results.

In addition to the alternations in the proportion of fiber type seen with age, gender and pathology, considerable variation is also seen with the location or depth within the muscle from which a particular muscle biopsy is sampled (Lexell, Henriksson-Larsen, &

Sjostrom, 1983; Lexell et al., 1988). For example, Dahmane et al (2005) found a predominance of Type II fibers at the surface and Type I fibers in the deeper regions of muscles in both upper and lower extremities.

6.4.1 Fiber Type Distribution

“Anatomical complexity is seldom fortuitous but develops in response to a functional demand. Thus in very few instances can one simple function be assigned to a particular muscle. Far more frequently, muscles fulfillboth a postural function involving

119 tonic activity and also participate in movements involving phasic activity” (Johnson et al., 1973). Tonic muscles have a higher percentage of Type I fibers and are capable of sustained muscle activity over prolonged periods. Phasic muscles on the other hand have a higher percentage of Type II fibers and tend to fatigue more quickly.

One of the primary functions of the lumbar muscles is to maintain an upright posture. Of the back muscles, LMT is thought to play an important role in lumbar stability, maintaining posture, and controlling shear forces. As such, it is not surprising that studies on cadaveric specimens have established that the LMT and the thoracic components of the erector spinae muscles have a higher percentage of Type I muscle fibers (Johnson et al., 1973; Fidler et al., 1975; Jowett et al., 1975; Sirca & Kostevc,

1985; Thorstensson & Carlson, 1987; Jorgensen et al., 1993; Rantanen et al., 1993;

Mannion, Dumas et al., 1997; Mannion, Weber et al., 1997).

Although the concentration of Type I fibers is greater in the lumbar back muscles, previous studies have found that the distribution of fiber types within a muscle to be heterogeneous, that is, differing significantly between regions (Elder et al., 1982). As mentioned previously, Dahmane et al., 2005 found that the distribution of muscle fiber types varies as a function of depth, with a predominance of Type II fibers at the surface and Type I fibers in the deeper regions of a muscle. If this is true, then regional variations in the size and fiber type distribution through the volume of a muscle may support the theory of “neuromuscular partitioning”. Neuromuscular compartments are architecturally distinct regions within a muscle that are independently innervated by separate nerve branches. These regions in turn have specific functional, task-oriented roles.

120

The current study found a predominance of Type I muscle fibers (60%) as well as a higher proportion of Type I fiber area (70%) for LMT as a whole in keeping with

LMT’s theorized tonic stabilizing role. Interestingly, the area occupied by Type I fibers was shown to increase from deep to superficial (Table 5.5), with the greatest difference in mean proportions observed between its deep and superficial regions (p < 0.05). There was no evidence of a difference between the deep and intermediate regions (p = .09) and no evidence of a difference between the superficial and intermediate regions (p = .47).

These results parallel those of Johnson et al (1973) which found relatively more Type I fibers in the superficial versus the deep regions of the lumbar erector spinae in a group of

6 post-mortem subjects. In direct contrast, Sirca and Kostevc (1985) found roughly the same proportion of Type I and II fibers in the lumbar paraspinal muscles (longissimus and multifidus), with slightly more Type I fibers in the deep regions of these muscles. In a recent study by Dickx et al (2009) histological differences between the deep and superficial regions of LMT were investigated using T2 relaxation times on MRI. Longer relaxation times have are associated with muscles having more Type I fibers (Dickx, et al., 2009). Fifteen health male subjects (mean age = 23.3 years) were investigated. The authors found significantly higher relaxation times in the deep versus the superficial region of LMT suggesting a higher percentage of Type I fibers in the deep region of

LMT. Since the use of MRI to investigate fiber type distribution in muscle is limited and requires further exploration, the results of this study cannot be considered conclusive.

The proportion of Type I and II muscle fibers has also been shown to vary systematically as a function of depth within other skeletal muscles (Lexell, Henriksson-

Larsen, & Sjostrom, 1983). Lexell and colleagues (Lexell, Henriksson-Larsen, &

121

Sjostrom, 1983; Lexell & Taylor, 1989) showed that the proportion of Type I fibers and the mean fiber area in the vastus lateralis muscle of young adult males was greater in the deep regions of the muscle than the superficial regions. With increasing age, this difference was much less marked and the systemic distribution of fiber types and fiber areas diminished becoming more homogeneous (Lexell et al., 1988; Lexell & Taylor,

1991).

The conflicting results between this and previous studies may be due to several factors. The small sample size used in this part of the study is clearly a limiting factor which could have affected the results. Since only one specimen was used, this portion of the study may be considered highly exploratory, and hence pilot in nature. It is without question that further research in needed to analyse regional variations in fiber type distribution through the volume of LMT using a larger sample size. In addition to small sample size, the results of the current study may also be a consequence of intrinsic factors related directly to the specimen such as age and functional adaptation.

In the current study, a greater proportion of Type I fiber area was found in the superficial region of LMT. Older individuals, as was the case of the specimen used here may require greater tonic muscle activity from LMT in order to facilitate the maintenance and control over posture. Selective loss of Type II fibers and fiber type transformation

(Type II to Type I) in response to chronic increase use and changing functional demands may have resulted in a higher proportion of tonic muscle fibers in all parts of LMT particularly the superficial region. Interestingly, the proportion of Type I fiber area within the deep region of LMT was greatest at L5 compared with all other levels, but was shown not to be statistically significant (p=0.076). This observation may reflect the need

122 to maintain optimal stability and the concurrent control of intersegmental movement particularly at the lumbosacral junction.

6.4.2 Fiber Size

Dubowitz et al (1985) report the average size of Type I fibers in human skeletal muscle to be in the range of 50-55µm (males and females), while that of Type II fibers to be 55-

60µm in males and 45-50µm for females. In regards to the back muscles, the diameter of

Type I fibers is within the range given by Dubowitz et al (1985), while data on the diameter of Type II fibers generally shows greater variability (Rantanen et al., 1994) .

In the current study the mean diameters of Type I and II fibers was 56.6µm and

45.4µm respectively which is similar in size to that reported previously. Our analyses showed that Type I fibers had larger diameters compared with Type II fibers (p<0.005) and that Type I fiber diameter increased as the spinal level increased from L1 to L5. This increase may be related in part to the increasing functional demands for stability at lower lumbar levels. Mannion et al (1997) proposed that “if an individual is not genetically endowed with an excess of Type I fibers (by number) then the muscle seems to adapt by modifying the relative size of the fibers types in an attempt to achieve the same end result in relation to fatigue resistance. An alteration of fiber size is more readily achievable than is a transformation from one fiber type to another (Goldspink, 1985)”.

As was stated previously, the current study found that the area occupied by Type I fibers (i,e. proportion of Type I fiber area) was relatively large for the deep region of

LMT at the L5 level. Mannion et al (1998) showed that there is a significant correlation between mean fiber size (i.e. distribution of fiber types and fiber diameter or CSA) and

123 electromyographic measures of back muscle fatigability. Hence, the larger diameter and proportion of Type I fiber area observed at more inferior levels of the lumbar spine suggests that this region fatigues less readily. This allows LMT to function tonically both to maintain posture and control intersegmental movement were it is needed most.

6.5 Functional considerations

It has been reported that the control and stabilization of the lumbar spineis mediated through the interaction and activity of several trunk muscles (Panjabi, 1992b,

1992a), with strong evidence demonstrating LMT in augmenting spinal stiffness (Panjabi et al., 1989; Kaigle et al., 1995; Wilke et al., 1995). Despite this evidence, several of the clinical beliefs regarding this muscle have not been substantiated (MacDonald et al.,

2006). The results from the current study challenges some previous theories concerning the role of LMT and the principles used therein to retrain this muscle using therapeutic exercise.

Detailed dissection and digitization of the entire LMT for ten specimens produced accurate 3D computer models of this muscle in situ using Autodesk® Maya® 2009. In particular, the deep region of LMT was observed to have morphological, architectural and fiber type characteristics which do not appear to support its role (except perhaps at the lumbosacral junction) in controlling lumbar segmental stability directly. Based on our results, deep LMT is best suited to provide proprioceptive feedback to the CNS.

Moseley et al (2002) demonstrated that the deep and superficial fibers of LMT are differentially active during single and repetitive movements of the arm. Specifically, they were able to show that the deep fibers of LMT became activated in a “non-direction-

124 specific manner to modulate spine compression for the control of intervertebral and rotational forces” (Moseley et al., 2002). However, the current study demonstrated that the size (i.e. volume and PCSA), architecture, and morphology of deep LMT to be inadequate to facilitate the control of intersegmental movement. Instead, the intermediate region of LMT, with its larger volume, PCSA, and complimentary muscle architectural seems better suited for controlling shear and/or torsion forces. Since the deep LMT becomes activated irregardless of the direction of limb movement this should not be interpreted as direct evidence in support of its theorized role in augmenting spinal stiffness. Instead, the non-direction specific activity observed in deep LMT may reflect its function to act as a specialized mechanoreceptor providing sensory information back to the CNS.

Since deep LMT is likely a spinal proprioceptor this region of the muscle should demonstrate low levels of continuous EMG activity. This is necessary in order for the

CNS to continuously monitor the “state of lumbar stability”. In addition, since it takes time to initiate a response based on sensory input, the CNS may generate an underlying level of tonic activity in deep LMT to increase the “readiness” of the spine to small changes in FBL. The ability to detect small changes in FBL is especially important around the neutral zone, where the spine exhibits the least amount of stiffness and hence where intersegmental movements between vertebrae are the greatest. The degree of readiness of deep LMT may be modulated by feedback (closed loop control system) from muscle spindles contained within its structure. It is the stretch reflex and its control of gamma motorneuron excitability that controls the sensitivity of the sensory component of muscle spindles.

125

From the preceding discussion, one might infer that deep LMT is likely to contain a high concentration of muscle spindles within its structure. Indeed, future studies would need to be conducted to verify this postulate. Interestingly, the results of this study demonstrated that the area occupied by Type I fibers was greater in the superficial versus the deep region of LMT. However, the gamma spindle system is known to facilitate the alpha motoneurons that control Type I muscle fibers (Johansson & Sojka, 1991). At first glance, the results of this study seem paradoxical. One explanation for this difference might be attributed to the age of the specimen used during the fiber typing portion of this study. Older specimens may have greater degenerative changes affecting the articular triad consisting of the intervertebral discs and zygapophyseal joints. As such, spinal stiffness increases with age decreasing the intersegmental nature of the spine. As the degrees of freedom available to the spine decreases so to will the need of the CNS to monitor the intersegmental movement between vertebrae. The proportion of Type I fibers within the deep region of LMT may consequently decrease thereby increasing relative proportion of Type I fibers observed in the superficial region of LMT.

The results of this study have important clinical implications. Firstly, clinicians may find themselves rethinking some aspects of how they retrain LMT, in particular the deep region in clients with LBP. The common practice of palpating deep LMT in patients with LBP for the purpose of isolating and hence isometrically contracting this region of the muscle should be reevaluated.

The results of this study suggest that all regions of LMT have an important role in promoting and maintaining spinal stability. Therefore to improve the effectiveness of therapeutic exercise programs designed torestore normal LMT function it is important to

126 optimize the activation and control over all of its regions concurrently whenever possible.

Therapeutic exercises that focus on the strength/endurance of LMT while concurrently facilitating its intersegmental stability and proprioception functionsmay improve clinical outcomes in patients with LBP. Further clinical research is necessary in this area.

Back surgery is performed routinely to relieve pain and improve quality of life in many patients with LBP (Chou, et al., 2009; Wilkinson, 1983). Unfortunately, in up to

10% of cases, surgery leaves the client no better and sometimes worse (Wilkinson, 1983).

A multitude of reasons may contribute to the relatively high rate of failed back surgery, including the patients’ age, sex, activity level, medical history, and socioeconomic status.

In addition, alterations to muscle architecture due to preexisting pathology do not spontaneously “correct” themselves following back surgery and may in fact be induced or even perpetuated with surgery. Hence, by restoring “normal” muscle architecture prior to back surgery it may be possible reduce recovery times. The current thesis has been able to document with cutting edge three-dimensional techniques the morphology and architecture of LMT throughout its volume and as such provides the prerequisite data with which to make comparisons with pathological muscle tissue. In addition, the use of real time ultrasound to retrain LMT in patients with LBP may be expanded to include the non-invasive visualization of its muscle architecture well as assists in fine wire electrode placement of specific regions of LMT to study their activation patterns. This may in turn aid clinicians in monitoring the effectiveness of specific therapeutic intervention strategies. Surgeons may find the data helpful when selecting patients most appropriate for surgery, and thus more likely to have a positive outcome.

127

Another possible reason for failed back surgery is that the innervation to LMT via the medial branch of the posterior ramus may be disrupted causing localized muscle atrophy and segmental instability. Minimal invasive surgery that does not disturb the neuromuscular arrangement and continuity of the back muscles is an integral part of achieving good post-operative outcomes. Not surprisingly, a thorough understanding of the spatial arrangement of both contractile (i.e. muscle fiber bundles) and non-contractile

(i.e. tendons, aponeuroses, and nerves) tissue elements is a necessary substrate in the formulation of new surgical techniques with the least iatrogenic effects.

6.6 3D Reconstruction and Modelling: Pros and Cons

The visualization and quantification of skeletal muscle architectural parameters is a worthwhile endeavor as it allows muscle function to be better assessed. As Macintosh and Bogduk (1986) correctly point “this can be done most accurately in cadavers in which all the fascicles of a muscle can be dissected, and their attachments visualized and plotted directly”. This process although capable of yielding accurate architectural data,

“is a time-consuming process if large numbers of observations are desired to cover biological variability” (Macintosh and Bogduk, 1986). Although more rapid approaches of determining muscle architecture using magnetic resonance imaging (MRI) and computerized tomography (CT) have been used, these methods cannot account for changes in fiber length and fiber angle that occur along the muscle length (Lieber &

Friden, 2000). In addition, although plotting the attachments of muscle fascicles on 2D clinical radiographs may be a more rapid approach to discerning the possible actions of a muscle, it is limited in that this method “collapses” the 3D architecture of muscles into a

128 few representative line vectors. Only rough approximations of muscle force can be generated using this approach.

Although 3D modeling and reconstruction of skeletal muscle in situ using digitization of individual fiber bundles is a more tedious and time consuming process, it has the advantage of providing a complete architectural data set. The models generated using this approach represents the closest approximations of the in situ characteristics of skeletal muscle and can be used to predict the gross (e.g. torque producing) and subtle

(e.g. control of intersegmental movement) actions of muscles. Finite element models may be produced with this data and in conjunction with specific tissue properties (e.g. tensile strength of tendons and ligaments and load tolerance limits of vertebrae and intervertebral discs). In addition, the ability to add or subtract individual muscle fibers, fascicles, or entire sections of muscle provides spinal biomechanists with a model to test the affects of muscle atrophy or dysfunction on the liability of different elements of the spine to injury. Finally, for educational purposes three dimensional reconstructions of muscle also permit clinicians, researchers and medical/allied health students to visualize muscle at a level of complexity and acuity that could not be achieved previously. These models may be developed further providing virtual surgery constructs to help plan and execute various surgical procedures.

Chapter 7 Conclusions

1. LMT is a highly complex muscle with a hierarchal arrangement of muscle fiber

bundles/fascicles which are grouped into architecturally distinct regions: superficial,

intermediate, anddeep.

2. Fascicles of LMT attaching to the L1 to L4 vertebrae contained superficial,

intermediate and deep regions, while fascicles attaching to the L5 vertebra contained

only superficial and deep regions. The intermediate region was absent at L5.

3. Mean FBL was significantly different between regions of LMT (p < 0.05), increasing

from deep (2.9 ± 1.1 cm) to superficial (5.8 ± 1.6 cm).

4. Mean FBA was not significantly different between regions, but showed an increasing

trend from superficial (13.7 ± 1.1°) to deep (18.3 ± 1.1°).

5. Mean volume and PCSA increased significantly (p< 0.005) from the deep to

superficial region.

6. The superficial, intermediate and deep regions of LMT were independently

innervated by branches of the medial branch of the posterior ramus.

7. The area occupied by Type I and Type II fibers was significantly different (p< 0.01)

between the deep and superficial regions of LMT possibly reflecting functional

diversification between these regions. In this pilot study using one fresh cadaveric

specimen, the superficial region contained 75% Type I fibers while the deep region

had only 56% Type I fibers.

129 130

7.1 Functional paradigm

LMT has neuromuscular compartments or regions based on muscle architecture, nerve distribution pattern and fiber type composition. The findings suggest that each region has a specific function, working together synergistically to control movement and/or provide stability to the lumbar spine.

Clinically, PCSA is proportional to torque production, while FBL is proportional to excursion (Lieber et al., 1992). In addition,asmall FBA relative to the long axis of the spine maximizes torque produced in the sagittal plane. The primary function of superficial LMT is posterior sagittal rotation (i.e. extension) of the lumbar spine due its long fiber bundles (5.8 ± 1.6 cm), small fiber bundle angles (13.7 ±6.9º), and large physiological cross sectional area (70.3 ±3.0 mm3). Within the superficial region, FBL decreased from L1 (7.3 ± 1.7 cm) to L5 (4.8 ± 1.7 cm). Muscle fiber bundles originating from upper lumbar vertebrae (e.g. L1, L2) may produce larger excursions and are likely more suitable in controlling and maintaining the lumbar lordosis. In contrast, fiber bundles originating from lower lumbar vertebrae (e.g. L4, L5) may produce smaller excursions and are better suited in restraining intersegmental movement. In principle, muscles with a greater proportion of Type I fibers fatigue less readily compared with those having a predominance of Type II fibers. In the current study the area occupied by

Type I fibers was greatest for the superficial region and is believed to be in response to increase use and changing functional demands placed on this region of LMT, particularly in older individuals. Superficial LMT may be tonically active in controlling anterior sagittal rotation (forward flexion) thereby helping to maintain the lumbar lordosis during forward bending.

131

Conversely, small FBL, large FBA and small PCSA is indicative of limited excursion range and torque production. Deep LMT is not well designed to produce torque or control intersegmental movement. Rather, it is well positioned to possibly function as a specialized mechanoreceptor providing proprioceptive feedback to the CNS. Indirectly, the relatively smaller proportion of Type I vs Type II fibers in this region does not support its proposed function in controlling intersegmental movement.

Intermediate LMT has architectural and histochemical properties which are

“intermediate” in value between the superficial and deep regions. Also, the placement of the tendon relative to the muscle belly suggests this region produces torque in a cranial to caudal direction and may facilitate the control of intersegmental movement. In addition, the absence of intermediate LMT at L5 may increase the incidence of certain pathological conditions, including spondylolisthesis or disc prolapse in certain older individuals.

Further study is needed to determine if the absence of intermediate LMT is a function of age and/or pathology.

As outlined above, each region of LMT has characteristic morphological, architectural, and histochemical properties which assist in ascribing specific roles to each region. The precise activation, timing and control over each of these parts by the central nervous system could be facilitated by the independent innervation observed to each region. Injury to one or more of these branches can lead to movement impairment syndromes and/or compromise stability. Great care is needed to preserve the nerve supply to all regions of LMT in patients undergoing surgery for LBP.

Chapter 8 Future Direction

The study undertaken for this thesis has many possible future directions, some of which are outlined below.

1. Develop ultrasound techniques to directly visualize and isolate different regions of

LMT for the purpose of electromyography, muscle biopsy and muscle retraining.

2. In light of new morphological, architectural, and nerve distribution data:

a. Re-evaluate clinical beliefs of the role of deep and superficial regions of LMT

in the control of stability and movement.

b. Develop new therapeutic treatment techniques to restore and/or optimize LMT

function in patients with LBP.

3. Explore the proprioceptor potentiality of deep LMT in motor control firstly through

direct investigation of muscle spindle distribution through the volume of LMT and

indirectly through studies which examine impaired proprioception in patients with

LBP.

4. Provide the impetus for further studies which examine the muscle architecture, nerve

innervation and fiber type distribution of other paraspinal muscles, such as those in

the cervical region which are commonly injured in whiplash.

5. Produce a dynamic 3D model of the entire musculoskeletal lumbar spine that is able

to predict the magnitude and direction of forces placed on it as a function of daily

activities, work or trauma.

132 133

6. Model pathologic muscle using architectural and ultrasound data collected from

patients with LBP, disc pathology or trauma, thus creating a tool to assess function

following surgical intervention.

7. Use the innervation data for LMT to:

a. Selectively target and denervate specific regions of LMT to examine its

effects on lumbar stability.

b. Improve radiofrequency neurotomy procedures involving the medial branch to

optimize therapeutic effects.

c. Improve surgical skill to lessen the trauma and improve outcomes after back

surgery.

Chapter 9 References

Adams, M. A. (2002).The biomechanics of back pain.Edinburgh ; New York: Churchill Livingstone.

Agur, A. M., Ng-Thow-Hing, V., Ball, K. A., Fiume, E., & McKee, N. H. (2003). Documentation and three-dimensional modelling of human soleus muscle architecture.Clin Anat, 16(4), 285-293.

Agur, A. M. R., & Grant, J. C. B. (2005). Grant's atlas of anatomy (11th ed.). Philadelphia: Lippincott Williams & Wilkins.

Alberts, B., Hunt, T., & Wilson, J. (1989). Molecular biology of the cell (2nd ed.). New York: Garland Pub.

Anderson, G. B. (1999). Epidemiological features of chronic low back pain.Lancet, 354, 581-587.

Asmussen, E., & Klausen, K. (1962). Form and function of the erect human spine.Clin Orthop, 25, 55-63.

Aspden, R. M. (1992). Review of the functional anatomy of the spinal ligaments and the lumbar erector spinae.Clinical Anatomy, 5, 372-387.

Bajek, S., Bobinac, D., Bajek, G., Vranic, T. S., Lah, B., & Dragojevic, D. M. (2000). Muscle fiber type distribution in multifidus muscle in cases of lumbar disc herniation. Acta Med Okayama, 54(6), 235-241.

Bergmark, A. (1989). Stability of the lumbar spine.A study in mechanical engineering.Acta Orthop Scand Suppl, 230, 1-54.

Biedermann, H. J., DeFoa, J. L., & Forrest, W. J. (1991). Muscle fibre directions of iliocostalis and multifidus: male-female differences. J Anat, 179, 163-167.

Boden, S. D., Riew, K. D., Yamaguchi, K., Branch, T. P., Schellinger, D., & Wiesel, S. W. (1996). Orientation of the lumbar facet joints: association with degenerative disc disease. J Bone Joint Surg Am, 78(3), 403-411.

Bogduk, N. (1983).The innervation of the lumbar spine.Spine, 8(3), 286-293.

Bogduk, N., & Endres, S. M. (2005). Clinical anatomy of the lumbar spine and sacrum (4th ed.). New York: Elsevier/Churchill Livingstone.

134 135

Bogduk, N., Macintosh, J. E., & Pearcy, M. J. (1992). A universal model of the lumbar back muscles in the upright position. Spine, 17(8), 897-913.

Bogduk, N., Wilson, A. S., & Tynan, W. (1982). The human lumbar dorsal rami.J Anat, 134(Pt 2), 383-397.

Brown, M. (1987). Change in fibre size, not number, in ageing skeletal muscle. Age Ageing, 16(4), 244-248.

Burkholder, T. J., Fingado, B., Baron, S., & Lieber, R. L. (1994).Relationship between muscle fiber types and sizes and muscle architectural properties in the mouse hindlimb.J Morphol, 221(2), 177-190.

Cholewicki, J., & McGill, S. M. (1996). Mechanical stability of the in vivo lumbar spine: implications for injury and chronic low back pain. Clin Biomech (Bristol, Avon), 11(1), 1-15.

Cholewicki, J., McGill, S. M., & Norman, R. W. (1991). Lumbar spine loads during the lifting of extremely heavy weights. Med Sci Sports Exerc, 23(10), 1179-1186.

Chou, R., Baisden, J., Carragee, E. J., Resnick, D. K., Shaffer, W. O., & Loeser, J. D. (2009). Surgery for low back pain: a review of the evidence for an American Pain Society Clinical Practice Guideline. Spine (Phila Pa 1976), 34(10), 1094-1109.

Chow, R. S., Medri, M. K., Martin, D. C., Leekam, R. N., Agur, A. M., & McKee, N. H. (2000). Sonographic studies of human soleus and gastrocnemius muscle architecture: gender variability. Eur J Appl Physiol, 82(3), 236-244.

Crisco, J. J., 3rd, & Panjabi, M. M. (1991). The intersegmental and multisegmental muscles of the lumbar spine.A biomechanical model comparing lateral stabilizing potential.Spine, 16(7), 793-799.

Cutts, A. (1988). Shrinkage of muscle fibres during the fixation of cadaveric tissue J. Anat, 160 75-78.

Dahmane, R., Djordjevic, S., Simunic, B., & Valencic, V. (2005). Spatial fiber type distribution in normal human muscle Histochemical and tensiomyographical evaluation.J Biomech, 38(12), 2451-2459.

De Foa, J. L., Forrest, W., & Biedermann, H. J. (1989). Muscle fibre direction of longissimus, iliocostalis and multifidus: landmark-derived reference lines. J Anat, 163, 243-247.

Delp, S. L., Suryanarayanan, S., Murray, W. M., Uhlir, J., & Triolo, R. J. (2001). Architecture of the rectus abdominis, quadratus lumborum, and erector spinae.J Biomech, 34(3), 371-375.

136

Dickx, N., Cagnie, B., Achten, E., Vandemaele, P., Parlevliet, T., & Danneels, L. (2009). Differentiation between deep and superficial fibers of the lumbar multifidus by magnetic resonance imaging. Eur Spine J.

Donisch, E. W., & Basmajian, J. V. (1972). Electromyography of deep back muscles in man. Am J Anat, 133(1), 25-36.

Dubowitz, V., Brooke, M. H., & Neville, H. E. (1985).Muscle biopsy: a practical approach (2nd ed.). London: Bailliere Tindall.

Elder, G. C., Bradbury, K., & Roberts, R. (1982).Variability of fiber type distributions within human muscles.J Appl Physiol, 53(6), 1473-1480.

English, A. W., & Letbetter, W. D. (1982a). Anatomy and innervation patterns of cat lateral gastrocnemius and plantaris muscles. Am J Anat, 164(1), 67-77.

English, A. W., & Letbetter, W. D. (1982b). A histochemical analysis of identified compartments of cat lateral gastrocnemius muscle. Anat Rec, 204(2), 123-130.

English, A. W., Wolf, S. L., & Segal, R. L. (1993). Compartmentalization of muscles and their motor nuclei: the partitioning hypothesis. Physical Therapy, 73(12), 857- 867.

Enoka, R. M. (1988). Neuromechanical basis of kinesiology. Champaign, Ill.: Human Kinetics Books.

Farfan, H. F. (1973). Mechanical disorders of the low back. Philadelphia: Lea & Febiger.

Farfan, H. F. (1975). Muscular mechanism of the lumbar spine and the position of power and efficiency.Orthop Clin North Am, 6(1), 135-144.

Fidler, M. W., Jowett, R. L., & Troup, J. D. (1975). Myosin ATPase activity in multifidus muscle from cases of lumbar spinal derangement.J Bone Joint Surg Br, 57(2), 220-227.

Ford, D., Bagnall, K. M., McFadden, K. D., Greenhill, B., & Raso, J. (1983). Analysis of vertebral muscle obtained during surgery for correction of a lumbar disc disorder. Acta Anat (Basel), 116(2), 152-157.

Friederich, J. A., & Brand, R. A. (1990). Muscle fiber architecture in the human lower limb.J Biomech, 23(1), 91-95.

Friedman, M. H. (1988).Functions of the medial pterygoid muscle.Clinicla Anatomy, 1(3), 213-220.

137

Gans, C. (1982). Fiber architecture and muscle function.Exerc Sport Sci Rev, 10, 160- 207.

Goel, V. K., & Gilbertson, L. G. (1995).Applications of the finite element method to thoracolumbar spinal research--past, present, and future.Spine, 20(15), 1719-1727.

Goel, V. K., Kong, W., Han, J. S., Weinstein, J. N., & Gilbertson, L. G. (1993). A combined finite element and optimization investigation of lumbar spine mechanics with and without muscles.Spine, 18(11), 1531-1541.

Goldspink, G. (1985). Malleability of the motor system: a comparative approach. J Exp Biol, 115, 375-391.

Gracovetsky, S., Farfan, H. F., & Lamy, C. (1977).A mathematical model of the lumbar spine using an optimized system to control muscles and ligaments.Orthop Clin North Am, 8(1), 135-153.

Grimby, G., & Saltin, B. (1983). The ageing muscle.Clin Physiol, 3(3), 209-218.

Guo, H. R., Tanaka, S., Halperin, W. E., & Cameron, L. L. (1999). Back pain prevalence in US industry and estimates of lost workdays. Am J Public Health, 89(7), 1029- 1035.

Hansen, L., de Zee, M., Rasmussen, J., Andersen, T. B., Wong, C., & Simonsen, E. B. (2006). Anatomy and biomechanics of the back muscles in the lumbar spine with reference to biomechanical modeling. Spine, 31(17), 1888-1899.

Hides, J. A., Jull, G. A., & Richardson, C. A. (2001). Long-term effects of specific stabilizing exercises for first-episode low back pain.Spine, 26(11), E243-248.

Hides, J. A., Richardson, C. A., & Jull, G. A. (1996). Multifidus muscle recovery is not automatic after resolution of acute, first-episode low back pain. Spine, 21(23), 2763-2769.

Hoffer, J. A., & Andreassen, S. (1981). Regulation of soleus muscle stiffness in premammillary cats: intrinsic and reflex components. J Neurophysiol, 45(2), 267- 285.

Johansson, H., Sjolander, P., & Sojka, P. (1991). Receptors in the knee joint ligaments and their role in the biomechanics of the joint. Crit Rev Biomed Eng, 18(5), 341- 368.

Johansson, H., & Sojka, P. (1991). Pathophysiological mechanisms involved in genesis and spread of muscular tension in occupational muscle pain and in chronic musculoskeletal pain syndromes: a hypothesis. Med Hypotheses, 35(3), 196-203.

138

Johnson, M. A., Polgar, J., Weightman, D., & Appleton, D. (1973). Data on the distribution of fibre types in thirty-six human muscles. An autopsy study.J Neurol Sci, 18(1), 111-129.

Jonsson, B. (1970). The functions of individual muscles in the lumbar part of the spinae muscle.Electromyography, 10(1), 5-21.

Jorgensen, K., Nicholaisen, T., & Kato, M. (1993).Muscle fiber distribution, capillary density, and enzymatic activities in the lumbar paravertebral muscles of young men.Significance for isometric endurance.Spine, 18(11), 1439-1450.

Jowett, R. L., Fidler, M. W., & Troup, J. D. (1975). Histochemical changes in the multifidus in mechanical derangements of the spine.Orthop Clin North Am, 6(1), 145-161.

Juniper, R. P. (1981).The superior pterygoid muscle?Br J Oral Surg, 19(2), 121-128.

Kaigle, A. M., Holm, S. H., & Hansson, T. H. (1995). Experimental instability in the lumbar spine.Spine, 20(4), 421-430.

Kim, C., Gottschalk, L., Eng, C., Ward, S., & Lieber, R. (2007).160. The Multifidus Muscle is the Strongest Stabilizer of the Lumbar Spine (Vol. 7): Elsevier Science.

Kim, S. Y., Boynton, E. L., Ravichandiran, K., Fung, L. Y., Bleakney, R., & Agur, A. M. (2007). Three-dimensional study of the musculotendinous architecture of supraspinatus and its functional correlations.Clin Anat, 20(6), 648-655.

Kippers, V., & Parker, A. W. (1984). Posture related to myoelectric silence of erectores spinae during trunk flexion. Spine, 9(7), 740-745.

Larsson, L., & Karlsson, J. (1978). Isometric and dynamic endurance as a function of age and skeletal muscle characteristics.Acta Physiol Scand, 104(2), 129-136.

Lee, P. J., Rogers, E. L., & Granata, K. P. (2006). Active trunk stiffness increases with co-contraction. Journal of Electromyography & Kinesiology, 16(1), 51-57.

Lexell, J., & Downham, D. (1992). What is the effect of ageing on type 2 muscle fibres? J Neurol Sci, 107(2), 250-251.

Lexell, J., Downham, D., & Sjostrom, M. (1986). Distribution of different fibre types in human skeletal muscles. Fibre type arrangement in m. vastus lateralis from three groups of healthy men between 15 and 83 years.J Neurol Sci, 72(2-3), 211-222.

Lexell, J., Henriksson-Larsen, K., & Sjostrom, M. (1983). Distribution of different fibre types in human skeletal muscles. 2. A study of cross-sections of whole m. vastus lateralis. Acta Physiol Scand, 117(1), 115-122.

139

Lexell, J., Henriksson-Larsen, K., Winblad, B., & Sjostrom, M. (1983). Distribution of different fiber types in human skeletal muscles: effects of aging studied in whole muscle cross sections. Muscle Nerve, 6(8), 588-595.

Lexell, J., & Taylor, C. C. (1989). Variability in muscle fibre areas in whole human quadriceps muscle. How much and why? Acta Physiol Scand, 136(4), 561-568.

Lexell, J., & Taylor, C. C. (1991). Variability in muscle fibre areas in whole human quadriceps muscle: effects of increasing age. J Anat, 174, 239-249.

Lexell, J., Taylor, C. C., & Sjostrom, M. (1988). What is the cause of the ageing atrophy? Total number, size and proportion of different fiber types studied in whole vastus lateralis muscle from 15- to 83-year-old men. J Neurol Sci, 84(2-3), 275-294.

Lieber, R., & Friden, J. (2000). Functional and clinical significance of skeletal muscle architecure.Muscle and Nerve, 23, 1647-1666.

Lieber, R. L. (2002). Skeletal muscle structure, function &plasticity : the physiological basis of rehabilitation (2nd ed.). Philadelphia: Lippincott Williams & Wilkins.

Lieber, R. L., & Blevins, F. T. (1989). Skeletal muscle architecture of the rabbit hindlimb: functional implications of muscle design. J Morphol, 199(1), 93-101.

Lieber, R. L., & Boakes, J. L. (1988). Muscle force and moment arm contributions to torque production in frog hindlimb. Am J Physiol, 254(6 Pt 1), C769-772.

Lieber, R. L., Fazeli, B. M., & Botte, M. J. (1990). Architecture of selected wrist flexor and extensor muscles.J Hand Surg [Am], 15(2), 244-250.

Lieber, R. L., & Friden, J. (2002). Morphologic and mechanical basis of delayed-onset muscle soreness.J Am Acad Orthop Surg, 10(1), 67-73.

Lieber, R. L., Jacobson, M. D., Fazeli, B. M., Abrams, R. A., & Botte, M. J. (1992). Architecture of selected muscles of the arm and forearm: anatomy and implications for tendon transfer. J Hand Surg [Am], 17(5), 787-798.

Lundon, K. (2003). Orthopedic rehabilitation science: principles for clinical management of nonmineralized connective tissue (p.34). Boston: Butterworth- Heinemann.

MacDonald, D. A., Moseley, G. L., & Hodges, P. W. (2006). The lumbar multifidus: does the evidence support clinical beliefs? Man Ther, 11(4), 254-263.

Macintosh, J. E., & Bogduk, N. (1986). The biomechanics of the lumbar multifidus.Clinical Biomechanics, 1, 205-213.

140

Macintosh, J. E., & Bogduk, N. (1991). The attachments of the lumbar erector spinae.Spine, 16(7), 783-792.

Macintosh, J. E., Pearcy, M. J., & Bogduk, N. (1993). The axial torque of the lumbar back muscles: torsion strength of the back muscles. Aust N Z J Surg, 63(3), 205- 212.

Macintosh, J. E., Valencia, F., Bogduk, N., & Munro, R. R. (1986).The morphology of the human lumbar multifidus.Clinical Biomechanics, 1, 196-204.

Maganaris, C. N., Baltzopoulos, V., & Sargeant, A. J. (1998). In vivo measurements of the triceps surae complex architecture in man: implications for muscle function. J Physiol, 512 ( Pt 2), 603-614.

Mannion, A. F. (1999). Fibre type characteristics and function of the human paraspinal muscles: normal values and changes in association with low back pain. J Electromyogr Kinesiol, 9(6), 363-377.

Mannion, A. F., Dumas, G. A., Cooper, R. G., Espinosa, F. J., Faris, M. W., & Stevenson, J. M. (1997). Muscle fibre size and type distribution in thoracic and lumbar regions of erector spinae in healthy subjects without low back pain: normal values and sex differences. J Anat, 190 ( Pt 4), 505-513.

Mannion, A. F., Dumas, G. A., Stevenson, J. M., & Cooper, R. G. (1998). The influence of muscle fiber size and type distribution on electromyographic measures of back muscle fatigability.Spine, 23(5), 576-584.

Mannion, A. F., Kaser, L., Weber, E., Rhyner, A., Dvorak, J., & Muntener, M. (2000). Influence of age and duration of symptoms on fibre type distribution and size of the back muscles in chronic low back pain patients.Eur Spine J, 9(4), 273-281.

Mannion, A. F., Weber, B. R., Dvorak, J., Grob, D., & Muntener, M. (1997). Fibre type characteristics of the lumbar paraspinal muscles in normal healthy subjects and in patients with low back pain. J Orthop Res, 15(6), 881-887.

Martin, D. C., Medri, M. K., Chow, R. S., Oxorn, V., Leekam, R. N., Agur, A. M., et al. (2001). Comparing human skeletal muscle architectural parameters of cadavers with in vivo ultrasonographic measurements. J Anat, 199(Pt 4), 429-434.

Mattila, M., Hurme, M., Alaranta, H., Paljarvi, L., Kalimo, H., Falck, B., et al. (1986). The multifidus muscle in patients with lumbar disc herniation. A histochemical and morphometric analysis of intraoperative biopsies.Spine, 11(7), 732-738.

McGill, S. (2007).Low back disorders : evidence-based prevention and rehabilitation (2nd ed.). Champaign, IL: Human Kinetics.

141

McGill, S. M. (1991).Kinetic potential of the lumbar trunk musculature about three orthogonal orthopaedic axes in extreme postures.Spine, 16(7), 809-815.

McGill, S. M. (2001). Low back stability: from formal description to issues for performance and rehabilitation. Exerc Sport Sci Rev, 29(1), 26-31.

McGill, S. M., & Norman, R. W. (1986). Partitioning of the L4-L5 dynamic moment into disc, ligamentous, and muscular components during lifting. Spine, 11(7), 666-678.

McGill, S. M., Patt, N., & Norman, R. W. (1988). Measurement of the trunk musculature of active males using CT scan radiography: implications for force and moment generating capacity about the L4/L5 joint. J Biomech, 21(4), 329-341.

McGill, S. M., Santaguida, L., & Stevens, J. (1993). Measurement of the trunk musculature from T5 to L5 using MRI scans of 15 young males corrected for muscle fibre orientation.8, 171-178.

Meier, M. P., Klein, M. P., Krebs, D., Grob, D., & Muntener, M. (1997). Fiber transformations in multifidus muscle of young patients with idiopathic scoliosis.Spine, 22(20), 2357-2364.

Mohagheghi, A. A., Khan, T., Meadows, T. H., Giannikas, K., Baltzopoulos, V., & Maganaris, C. N. (2007).Differences in gastrocnemius muscle architecture between the paretic and non-paretic legs in children with hemiplegic cerebral palsy.Clin Biomech (Bristol, Avon), 22(6), 718-724.

Morris, J. M., Benner, G., & Lucas, D. B. (1962). An electromyographic study of the intrinsic muscles of the back in man.J Anat, 96(Pt 4), 509-520.

Moseley, G. L., Hodges, P. W., & Gandevia, S. C. (2002). Deep and superficial fibers of the lumbar multifidus muscle are differentially active during voluntary arm movements. Spine (Phila Pa 1976), 27(2), E29-36.

Moseley, G. L., Hodges, P. W., & Gandevia, S. C. (2003). External perturbation of the trunk in standing humans differentially activates components of the medial back muscles. J Physiol, 547(Pt 2), 581-587.

Ng-Thow-Hing, V. (2001).Anatomically-based Models for Physical and Geometric Reconstruction of Humans and Other Animals.Unpublished PhD, University of Toronto, Toronto.

Norlund, A., & Waddell, G. (2000). Cost of Back Pain in OECD Countries. In A. Nacemson & E. Jonsson (Eds.), Neck and Back Pain. The Scientific Evidence of Causes, Diagnosis, and Treatment (pp. 421-425). Philadelphia: Lippincott Williams & Wilkins.

142

O'Sullivan, P. B., Phyty, G. D., Twomey, L. T., & Allison, G. T. (1997). Evaluation of specific stabilizing exercise in the treatment of chronic low back pain with radiologic diagnosis of spondylolysis or spondylolisthesis.Spine, 22(24), 2959- 2967.

Panjabi, M., Abumi, K., Duranceau, J., & Oxland, T. (1989).Spinal stability and intersegmental muscle forces.A biomechanical model.Spine, 14(2), 194-200.

Panjabi, M. M. (1992a).The stabilizing system of the spine. Part I. Function, dysfunction, adaptation, and enhancement. J Spinal Disord, 5(4), 383-389; discussion 397.

Panjabi, M. M. (1992b).The stabilizing system of the spine. Part II. Neutral zone and instability hypothesis.J Spinal Disord, 5(4), 390-396; discussion 397.

Pierobon-Bormioli, S., Sartore, S., Libera, L. D., Vitadello, M., & Schiaffino, S. (1981)."Fast" isomyosins and fiber types in mammalian skeletal muscle.J Histochem Cytochem, 29(10), 1179-1188.

Porter, M. M., Vandervoort, A. A., & Lexell, J. (1995). Aging of human muscle: structure, function and adaptability. Scand J Med Sci Sports, 5(3), 129-142.

Porterfield, J. A., & DeRosa, C. (1991). Mechanical low back pain : perspectives in functional anatomy. Philadelphia: Saunders.

Porterfield, J. A., & DeRosa, C. (1998). Mechanical low back pain : perspectives in functional anatomy (2nd ed.). Philadelphia: W.B. Saunders.

Powell, P. L., Roy, R. R., Kanim, P., Bello, M. A., & Edgerton, V. R. (1984). Predictability of skeletal muscle tension from architectural determinations in guinea pig hindlimbs.J Appl Physiol, 57(6), 1715-1721.

Rantanen, J., Hurme, M., Falck, B., Alaranta, H., Nykvist, F., Lehto, M., et al. (1993). The lumbar multifidus muscle five years after surgery for a lumbar intervertebral disc herniation. Spine, 18(5), 568-574.

Rantanen, J., Rissanen, A., & Kalimo, H. (1994). Lumbar muscle fiber size and type distribution in normal subjects.Eur Spine J, 3(6), 331-335.

Richardson, C., Hodges, P. W., & Hides, J. (2004). Therapeutic exercise for lumbopelvic stabilization : a motor control approach for the treatment and prevention of low back pain (2nd ed.). Edinburgh ; New York: Churchill Livingstone.

Richardson, C. A. (1999).Therapeutic exercise for spinal segmental stabilization in low back pain : scientific basis and clinical approach. Edinburgh ; Sydney: Churchill Livingstone.

143

Rosatelli, A. L., Ravichandiran, K., & Agur, A. M. (2008). Three-dimensional study of the musculotendinous architecture of lumbar multifidus and its functional implications.Clin Anat, 21(6), 539-546.

Roy, R. R., Bello, M. A., Powell, P. L., & Simpson, D. R. (1984). Architectural design and fiber-type distribution of the major elbow flexors and extensors of the monkey (cynomolgus).Am J Anat, 171(3), 285-293.

Roy, R. R., Bodine-Fowler, S. C., Kim, J., Haque, N., de Leon, D., Rudolph, W., et al. (1991). Architectural and fiber type distribution properties of selected rhesus leg muscles: feasibility of multiple independent biopsies. Acta Anat (Basel), 140(4), 350-356.

Roy, R. R., & Ishihara, A. (1997). Overview: functional implications of the design of skeletal muscles. Acta Anat (Basel), 159(2-3), 75-77.

Sacks, R. D., & Roy, R. R. (1982). Architecture of the hind limb muscles of cats: functional significance. J Morphol, 173(2), 185-195.

Sato, T., Akatsuka, H., Kito, K., Tokoro, Y., Tauchi, H., & Kato, K. (1984). Age changes in size and number of muscle fibers in human minor pectoral muscle. Mech Ageing Dev, 28(1), 99-109.

Sirca, A., & Kostevc, V. (1985). The fibre type composition of thoracic and lumbar paravertebral muscles in man. J Anat, 141, 131-137.

Steffen, R., Nolte, L. P., & Pingel, T. H. (1994). [Importance of the back muscles in rehabilitation of postoperative segmental lumbar instability--a biomechanical analysis]. Rehabilitation (Stuttg), 33(3), 164-170.

Thorstensson, A., & Carlson, H. (1987). Fibre types in human lumbar back muscles. Acta Physiol Scand, 131(2), 195-202.

Valencia, F. P., & Munro, R. R. (1985). An electromyographic study of the lumbar multifidus in man.Electromyogr Clin Neurophysiol, 25(4), 205-221. van Eijden, T. M., Koolstra, J. H., & Brugman, P. (1995). Architecture of the human pterygoid muscles.J Dent Res, 74(8), 1489-1495.

Verbout, A. J., Wintzen, A. R., & Linthorst, P. (1989). The distribution of slow and fast twitch fibers in the intrinsic back muscles. Clinical Anatomy, 2, 120-121.

Ward, S. R., Eng, C. M., Gottschalk, L. J., Kim, C. W., Garfin, S. R., & Lieber, R. L. (2006). The architectural design of the lumbar multifidus muscle supports its role as stabilizer. Journal of Biomechanics, 39(Supplement 1), S101-S101.

144

Ward, S. R., Hentzen, E. R., Smallwood, L. H., Eastlack, R. K., Burns, K. A., Fithian, D. C., et al. (2006). Rotator cuff muscle architecture: implications for glenohumeral stability. Clin Orthop Relat Res, 448, 157-163.

Ward, S. R., Kim, C. W., Eng, C. M., Gottschalk, L. J. t., Tomiya, A., Garfin, S. R., et al. (2009). Architectural analysis and intraoperative measurements demonstrate the unique design of the multifidus muscle for lumbar spine stability. J Bone Joint Surg Am, 91(1), 176-185.

Wickiewicz, T. L., Roy, R. R., Powell, P. L., & Edgerton, V. R. (1983). Muscle architecture of the human lower limb.Clin Orthop(179), 275-283.

Wilke, H. J., Wolf, S., Claes, L. E., Arand, M., & Wiesend, A. (1995). Stability increase of the lumbar spine with different muscle groups.A biomechanical in vitro study.Spine, 20(2), 192-198.

Wilkinson, H. A. (1983). The Role of Improper Surgery in the Etiology of the Failed Back Syndrome. In N. Wilkinson (Ed.), The failed back syndrome: Etiology and therapy (pp. pp 15-16). Philadelphia: Harper & Row.

Williams, P. (Ed.). (1995). Gray's Anatomy (38th ed.). Edinburgh: Churchill Livingstone.

Winer, B. J. (1971).Statistical principles in experimental design (2d ed.). New York,: McGraw-Hill.

Wood, W. W., Takada, K., & Hannam, A. G. (1986). The electromyographic activity of the inferior part of the human lateral pterygoid muscle during clenching and chewing.Arch Oral Biol, 31(4), 245-253.

Yamaguchi, G., Sawa, A., Moran, D., Fessler, M., & Winters, J. (1990 ). A survey of human musculotendon actuator parameters In I. J. M. W. a. S. L.-Y. W. (Eds.) (Ed.), Multiple muscle systems:Biomechanics and movement organization (pp. 717-773 ). New York Springer-Verlag.

Yoshihara, K., Shirai, Y., Nakayama, Y., & Uesaka, S. (2001). Histochemical changes in the multifidus muscle in patients with lumbar intervertebral disc herniation. Spine, 26(6), 622-626.

Zajac, F. E. (1989). Muscle and tendon: properties, models, scaling, and application to biomechanics and motor control. Crit Rev Biomed Eng, 17(4), 359-411.

Zhao, W. P., Kawaguchi, Y., Matsui, H., Kanamori, M., & Kimura, T. (2000). Histochemistry and morphology of the multifidus muscle in lumbar disc herniation: comparative study between diseased and normal sides. Spine, 25(17), 2191-2199.

145

Zoidl, G., Grifka, J., Boluki, D., Willburger, R. E., Zoidl, C., Kramer, J., et al. (2003). Molecular evidence for local denervation of paraspinal muscles in failed-back surgery/postdiscotomy syndrome. Clin Neuropathol, 22(2), 71-77.