Patient Population

This was a retrospective matched case–control study of 60 patients with HD managed at a single-institute over an 8-year-period from January 2012 to January 2020. All patients included in the study were diagnosed to have HD by the neurologists based on the criteria proposed by Hirayama et al.1 2 The criteria included weakness and wasting predominantly in the C7, C8, and T1 myotomes in one upper limb or asymmetrically in both upper limbs, insidious onset of symptoms in adolescents or adults in their early twenties, initial fast progression for 1 to 3 years, irregular coarse tremors in the fingers of the affected hand or hands, mild transient worsening of symptoms on exposure to the cold, electromyographic evidence of chronic denervation in the clinically or sub-clinically affected muscles, and an absence of objective sensory loss.

Since the currently accepted first line of management of HD is cervical immobilization with a collar, all the 60 patients were initially advised to wear a hard collar for 6 weeks. This was done to assess their compliance with this treatment modality. Patients not comfortable with the idea of long-term collar wear or those who were not compliant with collar-wear in the 6-week trial period were then referred for surgery.

The cases consisted of 30 patients who underwent CD, while the controls consisted of 30 age- and sex-matched patients with HD who were managed conservatively with long-term collar wear. The exclusion criteria for the study included history of cervical spine trauma or hand surgery, or the presence of any other concomitant spinal disease. Informed consent was obtained from all the patients prior to surgery. The study was approved by the institutional review board and has been reported as per the STROBE (STrengthening the Reporting of OBservational studies in Epidemiology) guidelines.

Surgical Procedure

A standardized procedure was followed for all the surgically managed patients. Following fiberoptic intubation, patients were positioned prone with the head secured in pins and neck in neutral position. Laminectomy was performed at the levels corresponding to the maximum dilatation of the cervical epidural venous plexus seen on flexion magnetic resonance imaging (MRI). Following laminectomy, the underlying epidural venous plexus was systematically sought for, coagulated, and resected. A midline dural opening was then made corresponding to the laminectomy levels, and an expansile duraplasty performed using fascia lata. Posterior fixation was not performed in any of the cases. The patients were ambulated the day after surgery with a cervical collar that they wore for 1 to 2 weeks.

Outcome Assessment

Patients in both groups were assessed clinically at a minimum follow-up of 6 months after conservative or surgical treatment. For patients in the surgical group, various radiological and electrophysiological characteristics were also assessed at the follow-up visits, and compared with the preoperative status.

Clinical outcome in all patients was assessed using two patient-reported outcome measures (PROMs)- the Odom's scale9 and a self-rated hand grip strength (SRHGS) score. PROMs are being increasingly used in outcome studies, and validated PROMs such as the Odom's scale10 are no longer considered a subjective measure, but a robust, objective assessment of the patient's functional status.11 While the Odom's scale has been previously used to assess outcome after surgery for HD, hand grip strength assessment has not been performed in previous studies on HD. We used five subjective categories for the SRHGS score to quantify the degree of patient-perceived change in their hand grip strength at follow-up:

0 = worsened, 1 = same, 2 = slightly better, 3 = moderately better, 4 = extremely better. Questions on self-rated grip strength have been previously found to correlate well with actual grip-strength values obtained on a dynamometer.12

Patients were categorized into two predefined subgroups based on their Odom's scale ratings at the last follow-up point. The “improvement” subgroup included patients with “excellent” and “good” results on the Odom's scale, while patients with “fair” or “poor” scores constituted the “no improvement” subgroup. Clinical “deterioration” was defined as a “poor” Odom's score and/ or a SRHGS score of 0.

Electrophysiological Assessment

Electromyography (EMG) and nerve conduction studies were performed in all patients to establish the diagnosis of HD. Thenar and hypothenar compound muscle action potentials (CMAPs) were recorded at the abductor pollicis brevis and abductor digiti minimi in response to stimulation of the median and ulnar nerves, respectively. The mean of the CMAPs at the wrist and elbow was calculated to arrive at the overall CMAP for each nerve. The mean overall ulnar/median (U/M) CMAP ratio was then calculated. For the analysis, values from the affected side were used in case of unilateral disease. For patients with bilateral disease, the amplitudes of the more affected hand were taken.

Radiological Evaluation

All patients underwent dynamic radiographs and gadolinium-enhanced cervical MRI in neutral and 45 degrees flexion. A uniform imaging protocol was used for all patients. The MRI measurement slab was manually positioned from the base of the occiput through the upper portion of the C7 vertebral body. The axial MR slices were aligned parallel to the C2-C3 discs space.

The following key radiological characteristics associated with a diagnosis of HD were looked for at presentation: lower cervical cord atrophy on neutral MRI, loss of attachment between the posterior dural sac and laminae, forward migration of the cord, and prominent enhancing epidural mass-like venous plexus on flexion MRI (Fig. 1). Localized cord atrophy was defined as a relative decrease in cord size when compared with the normal cord above and below the affected levels on sagittal and axial T2-weighted sequences, and was recorded as unilateral, bilateral or absent. The following additional radiological parameters were assessed by two independent observers (AAA and NR) preoperatively and at follow-up in patients who underwent surgery. Cord morphology was objectively assessed by measuring the sagittal anteroposterior diameter (APD) of the lower cervical cord (defined as the cervical cord from C4-C7 levels) at the mid-disc line formed by a straight line connecting the midpoint of the anterior and posterior edges of the disc spaces from C4-C5 to C7-T1 levels. The distance between the anterior and posterior edge of the spinal cord on the extension of this line was taken as the APD (Fig. 2A). To reduce individual differences, the mean lower cervical APD was taken relative to the APD at C2-C3 a level that was free of compression in all patients and hence used as an internal reference.

Close

Fig. 1 Sagittal magnetic resonance imaging demonstrating the key radiological characteristics of Hirayama disease: (A) Neutral T2 sagittal image demonstrating lower cervical cord atrophy; gadolinium-enhanced T1-weighted sequences in (B) neutral and (C) flexion demonstrating forward migration of the cord and prominent enhancing epidural mass-like venous plexus on flexion.

Export to PPT

Fig. 1 Sagittal magnetic resonance imaging demonstrating the key radiological characteristics of Hirayama disease: (A) Neutral T2 sagittal image demonstrating lower cervical cord atrophy; gadolinium-enhanced T1-weighted sequences in (B) neutral and (C) flexion demonstrating forward migration of the cord and prominent enhancing epidural mass-like venous plexus on flexion.

Close

Fig. 2 T2 sagittal magnetic resonance section demonstrating (A) measurement of the anteroposterior diameter (APD) of the lower cervical cord at the mid-disc levels from C4 to C7 levels. C2-C3 APD was taken as reference value for each patient; (B) flexion and (C) neutral magnetic resonance imaging showing forward migration of the lower cervical cord from C4-C5 to C7-T1 levels. “x” is the distance between the posterior edge of the spinal cord and posterior edge of the spinal canal and “y” is the distance between the anterior and posterior edges of the spinal canal. The forward migration of the cord on flexion was calculated as the difference between the x/y ratios in (B) and (C).

Export to PPT

Fig. 2 T2 sagittal magnetic resonance section demonstrating (A) measurement of the anteroposterior diameter (APD) of the lower cervical cord at the mid-disc levels from C4 to C7 levels. C2-C3 APD was taken as reference value for each patient; (B) flexion and (C) neutral magnetic resonance imaging showing forward migration of the lower cervical cord from C4-C5 to C7-T1 levels. “x” is the distance between the posterior edge of the spinal cord and posterior edge of the spinal canal and “y” is the distance between the anterior and posterior edges of the spinal canal. The forward migration of the cord on flexion was calculated as the difference between the x/y ratios in (B) and (C).

Forward migration of the lower cervical cord was measured on sagittal images at the mid-disc lines from C4-C5 to C7-T1 levels. The distance between the posterior edge of the spinal cord and posterior edge of the spinal canal was termed “x,” and the distance between the anterior and posterior edges of the spinal canal was termed “y” (Fig. 2B, C). The “x/y” ratio was thus the relative position of the posterior edge of the cord in the canal. The difference in this ratio in flexion and neutral positions was used as the indicator for the relative forward migration of the spinal cord in flexion. The mean forward migration of the lower cervical cord at various levels was used for analysis. Cervical alignment was measured on neutral radiographs by recording the C2-C7 Cobb's angle, that is, the angle between the lines parallel to the inferior endplates of C2 and C7 vertebral bodies. Instability was assessed on dynamic radiographs before surgery and at follow-up.

Statistical Analysis

The statistical software SPSS 17.0 (SPSS, Chicago, Illinois, United States) was used for analysis. Using a sample size of 30, an α value of 0.05, and an effect size of 0.8 used from a previous surgical series,13 the power of our study was calculated to be 0.9. Dichotomous data are described as counts and percentages, whereas continuous data are presented as means ± standard deviation. The unpaired t-test was used to compare preoperative and follow-up data for the whole cohort, and for the comparison of outcomes between the subgroups based on clinical improvement. A “p” value of less than 0.05 was considered significant.

Demographics and Clinicoradiological Characteristics

All the study subjects were males with a mean age of 20.39 ± 2.93 years with the duration of symptoms ranging from 14 to 48 months. The clinical and radiological characteristics of the cohort are listed in Table 1. A majority of the patients presented with bilateral, asymmetric symptoms, or signs. The levels of enhancement in the cervical cord varied from three to six, with C5-C7 being the commonest levels of involvement. Cord atrophy on axial MRI was noted in 52 patients, and was unilateral (Fig. 3A) in 33 (55%) patients and bilateral in 19 (32%) patients. The clinical and electrophysiological characteristics of the cases and controls were similar (Table 2).

Close

Fig. 3 Comparison of preoperative and follow-up magnetic resonance imaging of a 22-year-old man who presented with unilateral symptoms of Hirayama disease and underwent C3-C7 laminectomy. T2 axial images demonstrating (A) unilateral cord atrophy (arrow) at presentation and (B) cord expansion at follow-up 9 months after surgery; T2 sagittal images on flexion demonstrating (C) multiple flow voids from C3 to C7 and lower cervical cord compression at presentation, and (D) disappearance of the flow voids and absence of cord compression at follow-up; gadolinium-enhanced T1 sagittal images in flexion demonstrating (E) intense contrast enhancement in the dilated epidural venous plexus before surgery and (F) no residual enhancement at follow-up.

Export to PPT

Fig. 3 Comparison of preoperative and follow-up magnetic resonance imaging of a 22-year-old man who presented with unilateral symptoms of Hirayama disease and underwent C3-C7 laminectomy. T2 axial images demonstrating (A) unilateral cord atrophy (arrow) at presentation and (B) cord expansion at follow-up 9 months after surgery; T2 sagittal images on flexion demonstrating (C) multiple flow voids from C3 to C7 and lower cervical cord compression at presentation, and (D) disappearance of the flow voids and absence of cord compression at follow-up; gadolinium-enhanced T1 sagittal images in flexion demonstrating (E) intense contrast enhancement in the dilated epidural venous plexus before surgery and (F) no residual enhancement at follow-up.

Electrophysiological Findings

All patients had EMG evidence of chronic denervation in the C7-T1 myotomes. nerve conduction velocity studies for the entire cohort demonstrated less than normal mean CMAP values for the median and ulnar nerves (3.87 ± 0.50 mV and 3.51 ± 0.53 mV respectively). The mean U/M CMAP ratio was 0.88 ± 0.11. The results of the sensory nerve conduction tests were normal in all patients.

Clinical Outcomes in the Surgical Group

Follow-up clinical data was available in 25 (83%) patients in the surgical group. All patients in the surgical group underwent CD as outlined earlier. CD was performed at three levels in a majority (24/30) of the patients, and at more than three levels in the remaining few. A majority of the patients who underwent CD demonstrated clinical improvement at a mean follow-up of 29.61 ± 12.78 months (Table 3), with 15 patients reporting “excellent” outcomes on the Odom's scale. A majority of the patients reported a SRHGS score of 3 (moderately better hand grip) (Table 3). There was no difference in clinical improvement between patients with unilateral and bilateral, asymmetric disease (p = 0.34). MRIs of the three patients with clinical “deterioration” showed progressive atrophy of the lower cervical cord, and these patients were advised close follow-up.

Clinical Outcomes in the Control Group

Follow-up clinical data was available in 20 (66%) patients in the control group being managed with long-term collar wear. None of these patients reported clinical improvement on the Odom's scale; 18 (90%) remained status quo and 2 reported “poor” scores at a mean follow-up of 24.33 ± 9.38 months (Table 3). A majority of the patients reported a SRHGS score of 1 (no improvement in hand grip). Two patients with clinical “deterioration” refused surgical intervention and opted to continue conservative treatment.

Postduraplasty Radiological Changes

The cervical cord morphology at different levels relative to the C2-C3 APD is listed in Table 4. The mean lower cervical APD improved from a preoperative value of 0.74 ± 0.13 to 0.86 ± 0.21 (p = 0.01) at follow-up. Table 4 also lists the difference in “x/y” ratios in flexion and neutral position, denoting the forward migration of the cord, at different levels. Maximum cord migration occurred at C6-C7 level. Follow-up MRIs demonstrated a significant arrest of this flexion-induced cord migration (0.20 ± 0.11 and 0.03 ± 0.07 preoperatively and at follow-up respectively, p < 0.0001). This finding was accompanied by the disappearance of the enhancing venous plexus on flexion in all patients at follow-up (Fig. 3).

Subgroup analysis (based on clinical “improvement” and “no improvement” subgroups) (Fig. 4) in the surgical group revealed a similar decrease in forward cord migration in the two subgroups at follow-up. Patients in the “improvement” subgroup, however, demonstrated significantly better mean cervical APDs.

Close

Fig. 4 Bar-chart comparing postsurgical radiological and electrophysiological changes between subgroups with and without clinical improvement. APD, anteroposterior diameter; U/M CMAP, ulnar and median compound muscle action potential.

Export to PPT

Fig. 4 Bar-chart comparing postsurgical radiological and electrophysiological changes between subgroups with and without clinical improvement. APD, anteroposterior diameter; U/M CMAP, ulnar and median compound muscle action potential.

Alignment before Surgery and at Follow-Up

The mean preoperative C2-C7 Cobb's angle of 4.24 ± 3.88 degrees did not change significantly at a mid-term follow-up period of close to 30 months after surgery (2.19 ± 5.95 degrees, p = 0.13). There were no differences in alignment between the subgroups with and without clinical improvement (p = 0.42). None of the patients had evidence of instability on dynamic radiographs at presentation or at follow-up. There was no instance of symptomatic kyphosis at follow-up.

Postoperative Electrophysiological Changes

Postintervention analysis of the entire surgical group revealed no significant change in the median or ulnar nerve CMAPs (3.95 ± 0.58 mV and 3.73 ± 0.57 mV respectively), or in the mean U/M ratio (0.94 ± 0.20). Subgroup analysis (Fig. 4), however, yielded different results, with patients in the “improvement” subgroup demonstrating significantly better U/M CMAP ratios at follow-up than patients in the “no improvement” subgroup.

Complications

There were no perioperative surgical complications such as cerebrospinal fluid leakage or development of new neurological deficits. One patient had a superficial wound infection that was managed conservatively.

Pathogenesis of HD—Theories and Substrates

The flexion myelopathy of HD is believed to be related to repeated flexion-induced microtrauma in the anterior horn cells and ventral roots of the lower cervical cord. The ensuing ischemia in these neural elements manifests as the characteristic distal amyotrophy.14 While some of these pathological findings have been confirmed at autopsy and neuropathologic studies,2 15 the underlying mechanism of the flexion-induced cord damage remains a subject of debate.16 17

The most widely accepted theories for the pathogenesis of HD are dural-based. The first is that of an abnormally tight dura causing anterior displacement and compression of the cord on being stretched in flexion.17 This has been corroborated by the loss of elastin fibers and normal dural architecture in surgically resected specimens.18 19 Another theory suggests an overstretch of the dura on flexion occurring due to disparity between the growth of the vertebral column and the spinal content.20 Since males tend to have faster rates of vertebral column growth than females, this disproportionate-growth theory can also explain the male preponderance seen in HD.20

Non dural-based theories of HD include compression by anteriorly located osteophytic bars,17 flexion-induced venous stagnation,16 21 22 and underlying instability.6 23 The genesis of the latter hypothesis is related to the effectiveness of collar wear and reports of abnormal alignment24 and range of neck flexion25 rather than any definite radiological evidence of instability.

Conservative versus Surgical Management: Key Findings of Our Study

Conservative therapy of HD involves protracted collar-wear with or without the usage of neurotrophic drugs.26 Collar immobilization reportedly prevents the dynamic damage to the cervical motor neurons.6 This in turn stabilizes the clinical course of the disease, especially in patients with a shorter duration of illness.26 27 While collar-wear is considered the first line of management of HD, a recently established clinician-led consensus statement recommended surgical intervention in cases of relapse or a protracted course of the disease, or if the patient is uncomfortable with the idea of long-term collar wear.26

Our study, one of the few large case–control studies on the disease, demonstrated stabilization of the clinical status in almost all patients managed with long-term collar-wear. None of these patients, however, reported any clinical improvement, and two of them had worsened at follow-up. On the other hand, a majority of our patients managed with CD reported improvement of their symptoms at a similar follow-up period, indicating that the surgical procedure altered the clinical course of our patients to a greater degree than collar-wear did. It may be argued that varying disease severity at presentation could have contributed to this finding. This was, however, unlikely, given that both the groups were similar with respect to their duration of symptoms, clinicoradiological, and electrophysiological characteristics.

Changes in Cord Morphology after Duraplasty

Follow-up imaging of the surgical group demonstrated expansion of the lower cervical cord in almost all the patients. Similar morphological changes in the spinal cord are known to occur after decompression in conditions like cervical spondylotic myelopathy (CSM)28 29 and have been postulated to be reflective of improvement in axonal transport.29 Three patients in the surgical group reported progression of symptoms even after surgery. Repeat imaging in these patients ruled out instability or kyphosis and demonstrated progressive worsening of cord morphology. The cause of the clinical deterioration and cord changes remained unexplained.

Other Postsurgical Radiological Changes

The pathognomonic forward migration of the dura and cord associated with an intensely-enhancing epidural mass on flexion MRI in HD has been attributed to an expansion of the venous plexus between an overstretched dural sac and the posterior wall of the spinal canal.30 31 32 All our patients demonstrated complete radiological resolution of the phenomenon at follow-up.

There have been reports of abnormal cervical alignment and sagittal balance in patients with HD.24 25 33 These have not been documented to correlate with disease severity or clinical outcomes. In our study, CD following a laminectomy did not affect the overall cervical alignment at mid-term follow-up.

Electrophysiological Parameters and Postoperative Changes

The CMAPs of the ulnar and median nerves and the U/M CMAP ratio are decreased in HD.34 The latter has been found to be a more sensitive and specific index of the disease than the individual CMAP values.35 In our study also, both the individual CMAP values and the mean U/M CMAP ratio were reduced at presentation. The subgroup of patients with clinical improvement had significantly better ratios than those in the non-improvement subgroup.

Surgical Options in HD

Even though there is level III evidence to back the effectiveness of surgical intervention in HD,6 8 21 36 there is no clarity on the optimal surgical approach.26 The available surgical options include a variety of decompressive and fusion procedures, all of which attempt to address the pathology of a tight dural canal and subsequent flexion-induced cord compression. Fusion procedures, anterior or posterior, arrest segmental movement, and prevent the dural migration and cord compression during flexion. Duraplasty after laminectomy or laminoplasty,21 36 37 38 39 40 on the other hand, directly addresses the pathology of the tight dural canal. An added advantage of a posterior procedure is that it also addresses the flexion-induced dilatation of the epidural venous plexus seen in HD. This phenomenon is postulated to aggravate cord ischemia in the lower cervical region by its deleterious effects on the spinal cord microcirculation during flexion or Valsalva maneuvers.16 21

Selection of Surgical Procedure

A literature search revealed that while over 300 cases of HD6 8 26 were managed with anterior or posterior fusion, only 16 patients were treated with duraplasty without fusion (Table 5).16 21 36 37 38 39 40 Ours is the first large series that critically evaluates CD as a surgical option in HD.

Since all our patients had lordotic spines and none of them had any radiological evidence of instability, we chose not to subject any of them to a fusion procedure. Fusion sparing surgery would theoretically also allow for maintenance of cervical mobility—an important factor to consider in the management of the young population typically affected by HD. On the downside, there is an inherent risk of postoperative kyphosis developing after multilevel cervical laminectomy. This phenomenon has, however, not been demonstrated to affect functional outcomes at long-term follow-up.41 42 43

We chose to perform duraplasty via laminectomy in our patients. While laminoplasty could have been considered in the young study subjects undergoing multilevel cervical decompression, there is no evidence to indicate the superiority of laminoplasty over laminectomy in HD. The comparison between the two procedures in literature is mainly drawn from studies on CSM,44 a disease that is entirely different from HD in terms of pathogenesis, natural history, and the age profile of affected individuals. Further, literature also suggests that laminoplasty is not superior to laminectomy in patients without preexisting instability,44 45 with no observable differences with regard to functional outcomes.

Implications of Our Study

HD is labeled a self-limiting disease, with long-term collar-wear being one of the widely adopted modalities to arrest disease progression. With more than three-quarters of our operated patients experiencing clinical improvement, our study demonstrates that surgical intervention via CD not only arrests disease progression of the disease, but can even improve symptoms. CD uniformly halts forward dural migration and promotes cord expansion at follow-up (Fig. 3).

Importantly, the above gains are achieved without subjecting young patients to a fusion procedure with its mobility-related disadvantages; no significant alignment changes associated with the procedure were noted at mid-term follow-up. However, it may be cautioned that our average follow-up period of around 30 months may not have been sufficient to reveal substantial alignment changes. Larger scale prospective studies with longer follow-up will help to establish the utility of CD without fusion in HD.

Study Limitations

Our study has the inherent limitations of a retrospective analysis. Although the Odom's scale is a useful outcome tool, development and usage of a disease-specific scale for HD would provide a more robust outcome analysis. While our sample size was adequate, follow-up data could not be obtained in all the patients. A longer duration of follow-up would have provided more insight into the continuum of changes in cord morphology in HD, and may uncover potential post-laminectomy kyphotic changes requiring intervention.