Biomechanics or Necromechanics? Or How to Interpret Biomechanical Studies

Abstract

The field of biomechanics is inextricably linked with orthopaedic surgery: loads and load distribution play a major role in the problems we treat and in the success and failure of many of our treatments. Nonetheless, despite powerful investigational tools, I would argue biomechanics has made a relatively minor impact in clinical practice primarily because most studies fail to account for the major distinction between living and nonliving systems: adaptability. While any study requires a clear question or hypothesis or goal, without accounting for adaptability and tissue tolerance, these studies might well be termed "necromechanical." These studies will always have limited clinical relevance unless they contain several key features: 1.) A choice of a mechanical parameter which is arguably a surrogate for relevant biological behavior; 2.) A set of loading regimens which arguably represent the entire range of loadings experienced in vivo; 3.) An explicit discussion of tissue tolerance to the mechanical perturbations of the study; 4.) When appropriate (i.e., the question relates to longer-term effects), an explicit exploration of tissue adaptation over time. Without meeting these requirements, any biomechanical study is suspect and requires interpretation with great caution. When meeting these requirements, biomechanics can provide powerful tools to explain the function of the body and to predict the success or failure of treatments prior to using them on patients.

"When I use a word," Humpty Dumpty said, in a rather scornful tone, "it means just what I choose it to mean, neither more nor less."

"The question is," said Alice, "whether you can make words mean many different things."

"The question is," said Humpty Dumpty, "who is to be master - that's all."

Through the Looking Glass

Lewis Carroll

INTRODUCTION

"Biomechanics" as a discipline arguably arose out of the classic work of Giovanni Alfonso Borelli, who before his death in 1679, extensively explored the forces generated on and within human and animal bodies by the activities and functions of life.⁵ Borelli, however, was certainly not the first to recognize the relationship between the mechanical environment and animal responsiveness, as Galileo investigated such matters at least 40 years earlier, but was unfortunately in old age by the time he planned a new book, " De animalium motibus."² Therefore, it was Borelli's contribution which received far wider attention owing to its publication.

"Biomechanics" as a concept is very loosely interpreted and applied to many scientific investigations. I would argue that time has shown many, if not most of these investigations have limited relevance to biology and clinical medicine. Citation indices for most biomechanics-related journals are relatively low, suggesting that most articles are not frequently cited and by inference that they have not made a large impact. The Institute for Scientific Information (ISI)^* reports a maximum impact factor of 32 and a median of 2.1 for all 147 journals categorized as "cell biology." The 41 "biomedical engineering" journals (ISI uses no category, "biomechanics"), on the other hand, have a maximum impact factor of 3.47 and a median of 0.98. While impact factors are only one reflection of true impact, I suspect one of the reasons for the relatively low citation rates in biomedical engineering is the failure to consider the key aspect distinguishing living from non-living systems: biological responsiveness and tissue adaptation. Furthermore, perhaps these latter studies are cited within their own fields (an average of less than 1 time over the two years following publication according to the impact factor), but are not then cited in the related biological fields they are intended to impact. This raises a question of relevance.

In making various arguments about clinical relevance, I will briefly explore the historical derivation of "biomechanics" and suggest specific reasons why many investigations have not been relevant to biology, and why they never had the potential to do so. I will further propose criteria for reporting biomechanical studies which should be considered by both authors and readers attempting to insure relevance. First, however, it is critical to establish the notion that orthopaedic surgery has largely been an empirical, not scientific discipline.

ORTHOPAEDICS IS AN EMPIRICAL DISCIPLINE

Throughout most of recorded history the healing arts achieved advances through empiricism: healers observed what worked and what did not work, and tried new things which were more or less logical in terms of what was believed at the time. Only in the last part of the 20th Century did our theoretical knowledge of the way the body worked reach an adequately sophisticated stage where new treatments (e.g., new antibiotics to counter newly emerging resistant organisms) were introduced based upon knowledge gained through scientific study. It is fair to state surgical disciplines lagged behind those disciplines using primarily pharmacological approaches. Perhaps this is not surprising given the advances in cell and molecular biology and genetics: we know many cascades of sometimes complex events, and in many cases have developed methods to intervene in those cascades. The surgeons' approaches, on the other hand, remained largely empirical, rather than arising out of some fundamental understanding of the way the body worked. More importantly to the arguments I will pose, science in orthopaedic surgery has far more often been used to explain or support what we have already clinically observed than to accurately predict what new interventions will work based upon a fundamental understanding of biological responses. (Some exceptions are notable, such as bisphosphonates to reduce bone resorption.) This point alone would necessarily limit the clinical impact of science, and is as true of most scientific disciplines as it is of biomechanics. Thus, perhaps we should not be entirely surprised the contribution of biomechanics to clinical medicine is limited.

BIOMECHANICS

The Oxford English Dictionary¹ suggests the term "biomechanics" emanated from the Soviet theater. They quote Carter's monograph⁸ on the subject: "He (Meierhold) established a studio of bio-mechanics in1921. The principles of bio-mechanics . . . became systematically applied by him in the R.S.F.S.R. theatre in Petrograd from 1918 to 1922." The laws of bio-mechanics are founded on the study of the physiological construction of man. The system aims to produce men who understand the mechanism and laws of their structure and can, therefore, use it perfectly. It has established a principle of analysis by which each movement of the body can be differentiated and made fully expressive." By 1933, the 11th edition of Stedman's medical dictionary defined biomechanics as "the science of the action of forces, internal or external, on the living body" (the preceding 10th edition had no listing for "biomechanics").¹⁶ In his 1933 Presidential address to The American Orthopaedic Association, Arthur Steindler commented, "I visualize biomechanics as a powerful and indispensable ally of the orthopaedic clinician."¹⁷ He suggested the tools of mechanics were critical to grasping the way the musculoskeletal system functions in every day life and that the "mechanics of locomotion" was "still virgin soil."

Hatze¹² later proposed the following definition, "Biomechanics is the study of the structure and function of biological systems by means of the methods of mechanics." He further notes, "It would not be correct to state that 'Biomechanics is the study of the mechanical aspects of the structure and function of biological systems' because biological systems do not have mechanical aspects. They only have biomechanical aspects (otherwise mechanics, as it exists, would be sufficient to describe all phenomena which we now call biomechanical features of biological systems)." In making these distinctions, Hatze, as did Stedman's Dictionary, implied several important concepts: First, mechanics constituted merely a set of concepts and approaches to studying any system, living or not. Second, living systems had unique properties not possessed by non-living systems ("biological systems do not have mechanical aspects"). Third, the mere notion of "biological" (G. bios - life) implies a living system. It is these fundamental distinctions which appear worth exploring in some detail, for they have substantial implications in appropriately interpreting biomechanics literature.

What distinguishes living from dead systems? At the most fundamental level, a more or less arbitrary definition of "life" makes the distinction. Do strands of self replicating proteins (prions) reflect "life?" At the highest organism level, whether a comatose human being is alive or dead also depends upon an arbitrary definition, at times debated by ethicists or courts of law. However, for ordinary purposes, the distinction seems remarkably easier: living systems have the capacity to sense the environment, respond, and adapt over time. That is, they are not static, but rather through internal processes alter certain of their characteristics in response to external stimuli. Even this definition is not entirely without difficulties, since certain non-living mechanical systems and some computer systems "adapt" (i.e., artificial intelligence). But for practical purposes in those complex "biological" organisms and structures for which there would be no debate about life, the distinction appears sound. This being the case, "adaptation" seems to fulfill Hatze's implication of a unique property in living systems.

Biological systems respond to their external environment in a seemingly infinite number of ways, limited only by the interactions of the thousands of molecules constituting the system. In an individual, these interactions occur on many time scales, from nanoseconds to years, and decades. An adaptation might be, for example, the nearly instantaneous (milliseconds) muscular responses of a runner whose shoe sole unexpectedly strikes a small stone. The playing arm of a professional tennis player will enlarge over a period of years owing to extra use.¹¹

Consistent with the notion of adaptation, Black, in distinguishing the properties of living versus dead tissues commented that a living tissue could "repair itself, modify its behavior in both the short term and the long term, and in addition, possesses considerable esthetic qualities."³ He further commented, "There are real differences in the properties of tissues in vitro as compared to in vivo, and failure to recognize the transition between them is a major source of scatter in experiments in tissue mechanics." Thus, not only do living tissues adapt, but even their immediate tissue properties differ from those of the dead. Thus, we might modify Hatze's definition accordingly: Biomechanics is the application of mechanics principles to explore living (i.e., adaptable) systems.

NECROMECHANICS

If biomechanics entails the study of adaptable systems, is the study of non-adaptable (non-living) systems "necromechanics" (G. nekros - dead)? More importantly, how do we properly interpret studies of non-adaptable systems, or those which purport to study living systems, but ignore adaptation? Equally cogent in terms of interpreting experiments on dead tissues is Black's recognition of the differing properties of the living and dead.³ He suggested four "grades" of tissue for which properties would be increasingly different (and non physiological, and perhaps non-interpretable in biological terms): viable tissue in situ with no necrosis, viable tissue in vitro maintained in a suitable medium and at body temperatures, nonviable ("dead" cells) tissues maintained in some sort of medium and at body temperatures, nonviable tissue kept moist, but either dried or cooled at some time. Unfortunately, these latter sorts of tissues are the ones frequently used as a matter of pure convenience, yet they have differing properties. Researchers virtually never demonstrate whether those differences might have relevance to the question being raised. I suggest, however, in at least some cases they might well have relevance.

What sorts of studies might be considered necromechanical? A cadaveric study of fracture fixation comparing three implant constructs? A study of micromotion of four joint implants into artificial composite femurs? A finite element analysis determining the stress distribution in an arterial wall with fluid flow? A study of tendon strength in excised tendons? Yes, in each case if subsequent adaptation framework¹² would be purely "mechanical" ("necromechanical"?) not "biomechanical."

INTERPRETATION OF NECROMECHANICAL STUDIES

Can necromechanical studies have relevance or validity^*? The answer to that question depends entirely upon the question or hypothesis being addressed. If we presume a researcher wants to know whether one fracture implant construct is stronger than another and he or she is not interested in whether the clinical successes or failures of the implants are different, then the answer would be yes. However, one must then ask whether such information is relevant or meaningful in clinical terms. To a clinician (and more importantly to a patient), the strength of implant constructs per se is irrelevant. Whether or not the constructs are sufficient to result in successful treatment and avoid complications is relevant. A clinical result, though, depends not only upon initial fixation strength, but construct and tissue tolerance to loads applied and subsequent tissue adaptation. Further, from a clinical point of view, the question of which construct is stronger is not the right question: all might be sufficiently strong, or all inadequately strong, and in both cases the differences are moot.

The likelihood of relevance arises from several important principles: 1.) The mechanical parameter ^* must demonstrably relate to biological behavior; 2.) The mechanical parameter must be obtained with meaningful ("physiological") force magnitudes and directions; 3.) The short-term tolerances of the tissues to the parameter must be explicitly addressed; 4.) The long-term adaptability of the tissues to the parameter history, (e.g., strain history), must be explicitly addressed.

The first requirement for relevance relates to the choice of mechanical parameters, always surrogates for some biological behavior. The choice naturally depends upon the question being asked, but critically on the timescale of the question. One would likely choose different parameters if the time scale of the question were nearly instantaneous (e.g., acute fracture) in contrast to many years (e.g., long term joint implant loosening). It is incumbent upon a researcher to provide adequate arguments and rationale for their choice, and to link their choice to some biological behavior within a given time scale. In our examples, ultimate load to failure would be an appropriate mechanical parameter in a comparison of behaviors under a single load for different constructs. On the other hand, if one is interested in knowing whether one implant would likely lead to greater bone mass loss from stress shielding than another, ultimate load to failure would be irrelevant, and construct stiffness (under multiple loading regimens) would be more appropriate. Brown et al.⁶ demonstrated some mechanical parameters might relate to long-term bone adaptation, while others did not. The literature on the relationship between the mechanical environment and osteoarthrosis provides an example of failure to provide adequate arguments for choice of parameter. Most investigators report spatially peak contact stresses as a causative factor because it is intuitive, but nonetheless without explicit justification based upon data.^4,7,9,10 Maxian et al.,¹³ on the other hand, provided data suggesting the likelihood of cartilage degeneration in a dysplastic hip was related to contact stress/years, not some simple magnitude of stress. It seems intuitive that osteoarthrosis relates to repetitive loading, while it is far from clear a single contact stress could reflect those aspects of the load history critical to developing osteoarthrosis. Thus, it is incumbent upon authors to provide compelling arguments for their parameter, and these arguments must be based upon sound logic and plausible data.

The second requirement is the use of biologically meaningful force magnitudes and directions when ascertaining the parameter. Most often investigators use a magnitude and direction of convenience: an amount they can reasonably apply in an experimental system, and perhaps associated with a direction at some time of "peak" loading. This restriction does not typically apply to a numerical simulation, where large numbers of load scan be explored (and this is increasing with greater computational power). Interestingly, while seldom mentioned, and to my knowledge never systematically explored, loads believed to be physiological in magnitude when applied to cadaveric parts (bones, tendons, musculotendinous junctions) frequently result in construct failures or fractures.^** For this reason, loads are often applied below presumed peak load magnitudes. That does not mean sub-peak loads are not important, for as earlier noted, it is load history, and not merely a single load, which determines biological response or clinical outcome. However, any load magnitude must be justified in terms of the question.

The third requirement is an exploration of tissue tolerance. Tolerance in this context implies a relatively short time frame, prior to the material and structural alterations caused by biological adaptation. Obvious short term deleterious responses in our example of the fracture implant construct would include failure of the construct under a single or a few physiological load magnitudes, or even within the number of cycles of loading in the time frame prior to healing. (Total hip arthroplasty patients take approximately 1.1 million activity cycles per year¹⁴ or approximately 3000 steps per day, although with wide variations of about 15 fold.¹⁸) Most studies, however, apply a single load, and not the repetitive loadings which would be routinely experienced in vivo. (Even at the lowest level of 200 cycles per day, how many studies of fracture constructs report the numbers of cycles of loading required before the fracture is healed at some weeks or months?) Let us presume our study comparing two forms of fracture fixation demonstrates failure under one or even a few given static loading regimens with one of the implants, but not the other. Does that mean the latter is reasonable to use in patients? Not necessarily, because it might fail under repetitive loading, or under a loading regimen not studied. On the other hand, both might effectively work in vivo because either the load distribution or the tissue properties and tolerance are different in vivo. Thus, even for relative differences, one must exercise considerable caution interpreting meaning, particularly when the authors do not explicitly and appropriately explore tissue tolerance.

The fourth requirement is an explicit consideration of adaptation to the mechanical perturbations. Some adaptations will occur within seconds, since cells respond to mechanical perturbations within that time frame, but these will not necessarily result in clinically significant changes. Other adaptations, (e.g., the resorption of bone around a screw), may occur within weeks, and these changes may well affect clinical outcome. These adaptations may result in both geometrical and property alterations, so that the conditions at the beginning of the experiment (e.g., the security of fixation) no longer apply. The question of relevance then resides in whether the surgical constructs result in changes of some meaningful mechanical parameter leading to no adaptation, beneficial adaptation, or deleterious adaptation.

In this exploration, I have not emphasized the power of various biomechanical tools. While all biomechanical studies must be relevant to a living system, living systems are difficult to study because of normal biological variations and the complexity of interaction of many systems at many levels. Furthermore, we may not always have a proper biological surrogate for the behavior we wish to explore (just as we may not alway shave a proper biomechanical surrogate). That said, bench top biomechanical studies are typically more repeatable than in vivo experiments. Computational studies (e.g., finite element analysis) are not only absolutely repeatable (presuming mathematical stability), but also have the very powerful advantage that only the variable of interest can be altered while keeping everything else the same. This advantage cannot be entirely realized with bench top studies, let alone in vivo experiments. In a very real sense biomechanics studies can and should be complementary to clinical observations and experience.

The arguments I have posed suggest biological or clinical relevance demands several features. All scientific studies demand a clear question, hypothesis or goal, which either implicitly or explicitly notes a time frame. Biomechanical studies, however, have additional requirements: 1.) A choice of a justifiable mechanical parameter which is a surrogate for relevant biological behavior; 2.) A set of loading regimens which arguably represent the entire range of loadings experienced in vivo; 3.) An explicit discussion of tissue tolerance to the mechanical perturbations of the study; 4.) When appropriate (i.e., the question relates to longer-term effects), an explicit discussion of tissue adaptation over time. While these are necessary requirements, they are not always sufficient. Without meeting these requirements, any "biomechanical" study should be interpreted with great caution. When the requirements are met, on the other hand, biomechanics as a field offers an extremely powerful set of tools not only to explain how the bodyworks, but also predict the success or failure of treatments prior to attempting them in patients.