Arthroscopic treatment of scaphoid fractures and nonunions

Relevant anatomy and biomechanics

The scaphoid bone is the only carpal bone that bridges both the proximal and distal rows. As a result it is subjected to continuous shearing and bending forces. The scaphoid is tilted in 40 ± 3 degrees of palmar tilt degrees in the coronal plane with an intrascaphoid angle averaging and 32 ± 5 degrees in the sagittal plane from the central axis of the forearm. Heinzelmann et al. found that the scaphoid was most dense at the proximal pole where the trabecular bone is thickest and more tightly packed. The trabeculae were thinnest and more sparsely distributed at the scaphoid waist, which is where most fractures occur. They also showed that scaphoids in males were 4 mm longer than female specimens, and they were also significantly wider in their proximal pole. When considering operative fixation from an antegrade approach, the authors suggested that small screw sizes might be necessary for female patients as many of the commercially available standard screws are larger than the proximal pole of the female scaphoid.

Up to 80% of the scaphoid is covered by cartilage on five articulating surfaces. The entire proximal half of the scaphoid is an articular surface within the radiocarpal joint, which means that the blood supply can only enter on the palmar distal segments and dorsal ridge. In a study of 15 cadaver arms, Gelberman and Menon found that the primary blood supply to the scaphoid was from the radial artery and that 70% to 80% of the entire proximal pole’s blood supply is from branches of the radial artery entering along the dorsal ridge of the scaphoid along the scaphoid waist. Approximately 20% of scaphoids have only a single small perforating branch proximal to the waist, and 13% of scaphoids have no vascular perforations. For this reason proximal pole fractures may take 6 to 10 months to heal, and have a high incidence of avascular necrosis (AVN). The proximal pole is dependent on a single dominant intraosseous vessel that enters through the region of the tubercle; therefore the more proximal the fracture the more limited the blood supply, which increases the risk of nonunion and AVN. This was elegantly demonstrated in a study by Ramamurthy et al., who reviewed 159 patients with an established scaphoid nonunion following internal fixation and nonvascular bone grafting. They calculated a fragment ratio by dividing the proximal fragment length by the sum of the lengths of the proximal and distal fragments. After performing a multivariate analysis of the variables that influenced the union rate, only the site of the nonunion and the delay to surgery had a statistically significant effect. The more proximal the fracture (i.e., the smaller the fragment ratio) and the longer the delay before surgery, the lower the probability of union.

Most scaphoid fractures (70%–80%) occur at the midportion, or waist, whereas 10% to 20% occur at the proximal pole with a small percentage occurring at the distal pole. Weber and Chao showed experimentally that scaphoid waist fractures occurred with forced wrist hyperextension of more than 95 degrees combined with ulnar deviation, which resulted in impingement of the scaphoid waist on the dorsal rim of the radius. Fractures of the proximal pole of the scaphoid result from dorsal subluxation during forced hyperextension. Horii et al. studied 18 scaphoid fractures caused by punching, with the wrist in neutral to slight flexion. They postulated that the mechanism of the fracture was caused by the an axial load transmitted through the index metacarpal to the trapezium and trapezoid, which produced a shear stress at the waist of the scaphoid.

A number of factors predispose toward a nonunion. Because of the scaphoid’s narrow waist where the trabeculae are thinnest and are more sparsely distributed, fracture site displacement decreases the bony contact area for union. Any waist fracture with displacement of greater than 1 mm or angulation of greater than 15 degrees may lead to a nonunion if left untreated. Because the scaphoid is largely covered by cartilage, any fracture heals by intramembranous ossification, so there is no fracture callus to provide any initial stability. Scaphoid fractures heal by intramembranous ossification rather than callus formation. Premature wrist loading results in bending, shearing, and translational forces resulting in progressive distal pole flexion and pronation. Inadequate fracture site immobilization may lead to volar bone resorption as a response to the continued loading, which may culminate in a nonunion with a secondary humpback deformity.

Displacement of the fracture is a strong risk factor for delayed or failed union. Singh et al. performed a meta-analysis of 1401 scaphoids and showed that displaced fractures of the scaphoid have a four times higher risk of nonunion than undisplaced fractures when treated in a plaster cast. The duration of immobilization should be guided by progression of healing of the scaphoid fracture, but is typically 8 to 10 weeks. The time to union has been shown to be adversely affected by treatment delays. Mean union times reported by Mack et al. were 19 weeks for subacute scaphoid waist fractures and 28 weeks for subacute proximal pole fractures. Langhoff and Andersen reported similar union times in fractures delayed by 4 weeks or greater, with union times of 20.7 weeks for proximal pole fractures, 17.4 weeks for displaced waist fractures, and 12.5 weeks for undisplaced waist fractures. These older studies, however, relied only on plain radiographs to determine union. In a more recent study Grewal et al. studied 28 patients with isolated scaphoid fractures who were treated in a delayed fashion with a short-arm thumb spica cast. In their cohort, the observed nonunion rate was 17.9% with cast treatment when the diagnosis was delayed between 6 weeks and 6 months. The mean delay between time of injury and initiation of treatment was 10.5 ± 4.3 weeks (range, 6–23 wk). There were 7 proximal pole fractures, 20 scaphoid waist fractures, and 1 distal pole fracture. Union was defined as 50% bridging bone on a CT scan. Twenty-three fractures successfully united with casting alone, resulting in an 82% union rate. The mean length of time in the cast was 11.0 ± 6.5 weeks for the scaphoid waist fractures, and 14.2 ± 8.7 weeks for the proximal pole fractures. The single subject with a distal pole fracture was casted for 2.5 weeks. They believed that one of the reasons for the improved union rates and union times was a result of a more accurate assessment of bone healing using CT scans. Factors that were found to have a significant association with failure of cast treatment included the presence of diabetes (p = 0.03), fracture comminution (p = 0.05), and a humpback deformity (p = 0.02). If there is a nonunion with a humpback deformity, delayed surgical treatment also negatively affects the union rates. Euler et al. found that the incidence of persistent nonunion and the inability to correct the dorsal intercalated segmental instability (DISI) deformity were correlated with an increased delay between the time of fracture and time of volar wedge grafting.

Biomechanically, the longer the screw the more rigid the fixation, because longer screws reduce forces at the fracture site and spread bending forces along the screw. Dodds et al. performed a cadaveric study in which short screws or long screws were placed along the central scaphoid axis after an osteotomy was simulated at the waist. Scaphoids that were repaired with long screws were significantly stiffer than those repaired with short screws. In the clinical situation, when rigid fixation could not be provided by central screw placement alone (such as in extreme proximal pole fractures and nonunions), augmented fixation was performed by inserting a 0.062-inch K-wire or a mini headless screw from the distal scaphoid to the capitate. The notion of central screw placement has recently been challenged, however. Faucher et al. performed oblique scaphoid osteotomies along the dorsal sulcus in 8 matched pair of cadaver specimens to determine whether a screw placed perpendicular to the oblique fracture line would provide fixation strength that is comparable with that of a centrally placed screw. One scaphoid from each pair was randomized to receive a screw placed centrally down the long axis and a screw was placed perpendicular to the osteotomy in the other scaphoid. Each scaphoid underwent cyclic loading from 80 N to 120 N at 1 Hz until 2 mm of fracture displacement occurred or 4000 cycles was reached. The specimens that reached the 4000-cycle limit were then loaded to failure. They found no difference in number of cycles or load to failure between the two groups. They concluded that a perpendicularly placed screw provided equivalent strength to one placed along the central axis. Screws placed perpendicular to the fracture line were also significantly shorter than screws placed down the central axis.

Diagnosis

After an acute fracture, patients will typically present with complaints of radial-sided wrist pain and hand weakness due to a fall onto a hyperextended wrist. They will be tender over the snuffbox, have a limited range of wrist motion, and may have a hematoma over the anatomic snuffbox. They may have a painful Watson test. These findings may be mimicked by a scapholunate (SL) ligament injury, however. After a chronic injury there may be minimal findings, save for complaints of radial-sided wrist pain with forced wrist extension during a pushup position, or resisted torqueing such as turning a doorknob.

Standard radiographs of the scaphoid include a posteroanterior (PA) view with the wrist in ulnar deviation, a lateral view, a semipronated view, and a semisupinated view. Unless the x-ray beam lies in the same plane as the fracture, the fracture line may be missed. The incidence of a false-negative radiograph is between 2% and 25%. Because failure to treat a stable scaphoid fracture within 4 weeks increases the nonunion rate, all clinically suspected scaphoid fractures are treated with immobilization until the cause of the symptoms is clarified. Follow-up radiographs and clinical examination are performed at 2 weeks. If a fracture is still suspected in the presence of negative repeat radiographs, an MRI is the most reliable imaging modality for detecting acute and occult fractures and is generally diagnostic within 24 hours of injury. A CT scan can provide additional information about the architecture or displacement of the fracture and for guidance in treatment. Adey et al. recommend that it should be used with caution for triage of nondisplaced scaphoid fractures because false-positive results occur. They believe that a CT is better for ruling out a fracture than for ruling one in. MRI is also helpful for evaluating scaphoid nonunions and evaluating the presence or absence of AVN. MRI evidence of AVN is based on the loss of normal T1 signal intensity of marrow fat in the proximal pole of the scaphoid. In cases of established scaphoid fracture with nonunion or malunion, a CT scan is helpful to define the anatomic details of the fracture and for planning operative intervention.

AVN of the proximal pole is a significant risk factor for nonunion. In a now classic paper, Green noted that the absence of punctate bleeding in the cancellous bone of the proximal pole at the time of surgery is a better predictor of AVN than the appearance of the preoperative radiographs. Lutsky pointed out that intraoperative punctate bleeding is subjective and may not be a reliable and accurate reference standard for AVN of the proximal pole. Biopsy can be subject to sampling error, given that AVN can be patchy. Megerle et al. compared preoperative contrast-enhanced MRI to assess the intraoperative bleeding of the proximal fragment in 49 patients and found that diminished or absent vascularity was predicted with a specificity of 90% by a preoperative contrast-enhanced MRI. Schmitt et al. studied the use of contrast-enhanced MRI with intravenous gadolinium compared with nonenhanced MRI in 88 patients who underwent surgical treatment for a proximal pole nonunion. They then graded the osseous viability by means of the number bleeding points. They found that the sensitivity for detecting avascular proximal fragments was significantly better (p <0.001) in contrast-enhanced MRI compared with nonenhanced MRI. Smith et al. investigated the use of a preoperative and postoperative CT scan in 31 patients who underwent an ORIF and bone grafting of a scaphoid nonunion and compared this to the histological findings. Of the various CT parameters measured, increased radiodensity of the proximal pole was found to have the strongest correlation with AVN (p <.004), with all 12 cases having histologically proven AVN. The increased radiodensity of the proximal pole (p <.05) also statistically correlated with the postoperative union rates.

Treatment

Surgical treatment

Most scaphoid screws can be inserted percutaneously or through a miniopen approach. There are some instances, however, where arthroscopic assistance can be useful. For instance, it can guide the starting point for guide wire placement in the proximal pole with dorsal insertion. It is also a valuable aid to assess the quality of the reduction, to guard against screw cut-out, and to evaluate the rigidity of fixation as seemingly good screw purchase may not adequately stabilize a comminuted segment.