Disorders of posture, balance, and gait associated with falls and related injuries are among the most debilitating symptoms of advancing Parkinson's disease (PD). In his seminal studies of these clinical sequelae in patients with postencephalitic Parkinsonism entitled The Basal Ganglia and Posture published over forty years ago, the British neurologist James Purdon Martin documented with great detail the disorders of postural fixation, righting reactions, and locomotion that similarly often accompany the progression of idiopathic PD [¹]. His observations on facilitating functional movements—by gently rocking patients prior to chair rising or gait initiation, by placing bold transverse lines on the walking surface, or through the use of vision to compensate for proprioceptive deficits—influenced the development of current rehabilitation approaches. Purdon Martin later summarized his perspectives on the integration of posture and movement by emphasizing that “…postural activity-should be regarded as a function in its own right and not merely as a component of movement…” [²]. He concluded that all of the conditions on which stepping and consequently locomotion depend (e.g., antigravity support of the body, equilibrium, propulsion) are postural in nature and that even stepping, which prevents the body from falling forward, serves a postural function.

Although neurophysiological studies in both quadrupeds and humans have indicated that the control of posture and locomotion is interdependent at many levels of the central nervous system (CNS) encompassing multiple supraspinal and spinal networks [³–⁶], the ways by which locomotion may be affected by prevailing postural conditions are not well understood. This is particularly relevant to the problems of postural instability and gait disorders that accompany advancing PD. Such problems are especially evident when patients with PD attempt to perform complex whole body posture and locomotion sequences during functional activities such as gait initiation, sit-to-stand and other transfers, turning while standing and walking, and in ongoing gait. During such tasks, hesitation delays or “freezing” episodes are frequently observed, and the normal pattern of spatial and temporal sequencing between postural and locomotor elements of the task is either absent or disrupted [⁷]. Thus, the question arises as to whether or not at least some of the difficulties with locomotion experienced by individuals with PD are attributable to a dysfunction of the neuronal networks that mediate the coupling between posture and locomotion. This issue appears to have important implications for current rehabilitative interventions. For example, current physical therapy and rehabilitation interventions for posture, balance, and gait disorders in PD mainly focus on separate aspects of the problems such as posture and balance training [⁸, ⁹] or gait training [¹⁰–¹⁴]. However, impaired coupling between posture and locomotion could contribute to gait and mobility disorders, due not only to biomechanical limitations but also to adaptive changes in neural control. For example, De Nunzio et al. have recently demonstrated that alternate rhythmic vibration during quiet stance of bilateral paraspinal muscles affecting trunk posture produced a cyclic transfer of the center of pressure mimicking the one accompanying body progression during walking [¹⁵]. When vibration was applied to the trunk musculature during gait, walking velocity, cadence, and stride length increased in both patients with PD and controls [¹⁶]. In contrast, no effects on gait were observed when leg muscles (soleus and tibialis anterior) were similarly vibrated. Since the paraspinal muscles contralateral to the single support stance leg play a role in the stabilization of trunk posture during stance, these results suggest that proprioceptive feedback from postural muscles can be used to improve the coupling of posture and locomotion elements of the gait cycle, thus facilitating performance of the task in people with PD.

The purpose of this perspective and review is to present a framework with supportive research findings to advance the viewpoint that focusing rehabilitation interventions on individual or isolated components of posture, balance, or gait disorders in persons with PD should be reevaluated. Instead, it is argued that the emphasis in intervention approaches ought to be shifted towards therapeutic training programs that directly target impairments in posture and locomotion coupling as a causal factor contributing to balance and gait dysfunction.

The difficulties with performing complex whole body posture and locomotion sequences during functional activities such as gait initiation [¹⁷–²¹], turning [²², ²³], sit-to-stand [²⁴, ²⁵], and ongoing walking [¹⁶, ²⁶] are commonly accompanied by timing delays in the coupling between postural movements of the body segments and the goal-intended locomotion action (e.g., stepping release, step redirection change in turning while walking, seat-off in chair rise, continuous walking). The conceptual and theoretical framework for developing intervention approaches that target impairments in posture and locomotion coupling is illustrated by focusing on the initiation of gait. During gait initiation, an anticipatory postural adjustment (APA) phase normally precedes and accompanies the initiation of the stepping phase [²⁷–³⁰]. For forward stepping, these APAs involve a sequence of muscle activations and changes in the ground reaction forces (loading of the initial swing leg and unloading of the initial stance leg) that move the net center of pressure beneath the feet backward and toward the initial swing limb. This motor sequence, which ends after heel off, produces the forces and moments necessary to propel the body center of mass (COM) forward and towards the single stance limb prior to stepping.

Compared with healthy control subjects, the mediolateral (M-L) and anteroposterior (A-P) ground forces characterizing APAs in patients with PD are abnormally prolonged in duration and reduced in amplitude with a delay in the sequencing between the beginning of the APA and step onset [¹⁷, ¹⁸, ³¹, ³²]. This delay may include abnormal pauses that disrupt the posture-movement coordination and may precipitate freezing of gait (FOG). While the APA is normally almost always present during voluntary stepping, it may often be absent in patients with PD [¹⁷, ¹⁹, ²⁰]. In such cases, hesitation delays are readily observable. Thus, the normal spatial and temporal coordination between the APA and stepping components of gait initiation is disrupted in PD in association with start hesitation and FOG.

In gait initiation, the anticipatory nature of the postural-step coordination appears to involve a role for motor prediction. A forward internal model (Figure 1) is a neural mechanism that predicts (estimates) the future state of a system given the current (actual) state and the sensorimotor control signals [³³–³⁷]. The use of a forward model for coordination between posture and locomotion could operate such that the neural circuits for initiating stepping would normally be actively delayed until the APAs that generate the weight transfer from bipedal to single leg support have achieved single stance limb loading [³⁸, ³⁹]. This transition in stance support reflects a change in the body center of mass-base of support (COM-BOS) relationship. Thus, using internal and external feedback information, the forward model would determine if the APAs have achieved the sufficient anticipated postural state (e.g., COM position and motion relative to the BOS) before initiating the gait cycle and finishing the postural phase.

In Figure 1, the integrated networks for posture and locomotion are activated in parallel [³⁸–⁴¹] to generate a posture command for segmental orientation and balance and a step command. These motor outputs will modify the COM-BOS relationship. If an external mechanical or sensory event that assists with the APA by facilitating weight transfer is applied early in the postural adjustment phase, sensory information about the limb loading conditions, together with an efference copy of the motor commands sent to the forward model estimating the anticipated limb loading conditions, can be used by the CNS to modify the two commands in advance based on an internal representation of the body. Sensory information produced by movement can also be used online to modulate posture and movement via external feedback mechanisms. Conceivably, the posture assistance provided by external mechanical effects and/or sensorimotor enhancement could decrease the completion time of the weight transfer compared with the predicted time of completion without assistance and/or improve the fidelity of the information associated with the changes in limb loading reflecting postural state conditions during the APA. Based on a mismatch between the predicted and actual limb loading conditions determined from the forward internal model, the initiation release of stepping would be advanced in time and occur earlier. Reinforcement of the posture-locomotion coordination with posture-assisted locomotion (PAL) training could lead to adaptive changes in the internal model for step initiation.

The mechanisms contributing to impaired gait initiation in PD are poorly understood. It has been hypothesized that postural instability and gait dysfunction in PD result from alterations in the output of the basal ganglia to the pedunculopontine nucleus (PPN) in conjunction with the progressive degeneration of the large cholinergic neurons of this nucleus [⁴², ⁴³]. The PPN has important inputs to regions of the mesencephalic extrapyramidal area and pontomedullary reticular formation that play a role in the pattern generation for locomotion and integration of posture and movement [⁴⁴]. Alternatively, it has been proposed that impaired gait initiation results from dysfunction of the basal ganglia and a resulting suppression or underactivity of the supplementary motor area [⁴⁵–⁴⁷], a region of the frontal cortex critically involved in the planning and preparation for movement. Models of posture and movement coupling [³⁸, ⁴⁰], such as the model presented in Figure 1, often emphasize that the voluntary command to initiate movement, including the timing signal, must be integrated with brain stem and spinal centers that mediate the control of posture. Accordingly, the supplementary motor area may play a role in providing feedforward information about the internal model to both the basal ganglia and posture and locomotion control regions in the brain stem. The fact that levodopa can often improve gait initiation and locomotion in patients with off-medication impairment [⁴⁸] provides evidence that alterations in basal ganglia output to both cortex and brain stem likely play a role in both the triggering of movement initiation and coupling of posture and locomotion. However, in advanced disease, posture and gait abnormalities often become resistant to levodopa replacement therapy, suggesting that the progressive degeneration or dysfunction of nondopaminergic regions of the neuraxis [⁴⁹], such as the PPN, becomes the principal pathology that mediates the disordered coupling between posture and locomotion.

In PD, difficulties with achieving the postural prerequisites for stepping could contribute to gait initiation delays, “start hesitation,” and FOG. With postural assistance, the usually prolonged APA duration and reduced amplitude accompanying PD could be, respectively, shortened and increased to enhance posture requirements and allow an earlier step onset time. Thus, the rationale for the PAL training approach is that the expected limb loading conditions associated with weight transfer to the single stance limb are enhanced (e.g., achieved earlier and more effectively) compared with what would usually be expected without the assistance. If patients with PD retain the capacity to adapt their putative internal model for stepping with PAL training, then it might be possible to remodel the timing sequence and other characteristics of posture and locomotion components of gait initiation.

In this perspective and review, we have advanced the view point that approaches to rehabilitation interventions that focus on changing separate isolated components for posture, balance, and gait in persons with PD may have limited effectiveness due to the importance of posture and locomotion coupling. Alternatively, there is neurophysiological and experimental support for the idea that posture and locomotion are highly integrated components of action that require understanding of how these control functions are interactively coupled. Moreover, there is evidence to indicate that individuals with PD have particular problems with coupling or sequencing posture and locomotion during complex whole body movements that are associated with falls. Expanding or shifting current conceptual and theoretical models of rehabilitation beyond posture/balance and gait-centered intervention focuses by incorporating posture and locomotion coupling problems as a target for rehabilitation outcomes may help to optimize and improve the effectiveness of current clinical practice in this important area of rehabilitation for persons with PD.

Some of the work reported in this paper was supported by the US National Institutes of Health Grant R21HD055386. The contributions of Yunhui Zhang, M.S., Tanya Simuni, M.D., Mario Inacio, M.S., Judith Morgia, B.S., and Lisa Shulman, M.D., are gratefully acknowledged.