The management of scaphoid fracture and nonunion is still challenging for hand surgeons. If not treated properly, it will lead to considerable consequences in patients' wrist function. The goal of treatment is to achieve union, correct the deformity, relieve pain, improve function, and prevent the progression of osteoarthritis [2, 3]. In our study, the arthroscopic treatment of displaced scaphoid fracture and nonunion using two-HCS fixation and a distal radius bone graft has provided satisfactory results, which were composed of a union rate of 100%, and significantly improved clinical outcomes.
The difference in union rates for conservative treatment of scaphoid fractures may be due to the combination of heterogeneous patient groups with different fracture types and locations. Gellman et al.  reported a union rate of greater than 95% of nondisplaced scaphoid fractures treated with casting immobilization. Bond et al. showed a 100% union rate of nondisplaced scaphoid fractures with casting immobilization, but longer radiographic healing time compared with surgical fixation. Several other comparative studies in patients with acute scaphoid waist fractures demonstrated better ROM of the wrist, grip strength, faster healing time, and earlier return to work with surgical fixation [20, 21]. A meta-analysis involving 378 patients with nondisplaced or minimally displaced scaphoid fractures reported that surgical fixation could result in decreased rates of delayed union, and decreased time away from work . However, another meta-analysis documented that patients treated surgically had an odds ratio of 6.96 for complications compared with patients treated nonsurgically . The long-term risks and benefits of surgical intervention should be carefully weighed and considered when managing the scaphoid waist fractures.
Biomechanical studies demonstrated that multidirectional movement of the scaphoid generated bending, rotational and translational forces at the fracture site, thereby exerting a negative impact on bone healing and giving rise to fibrous tissue formation [24, 25]. Rigid fixation was definitely of great importance to resist these forces and facilitate primary bone healing. Reliable union rates and good functional outcomes were reported in patients treated with volar plate fixation [26, 27]. However, it was suggested that this technique has significant complications such as plate breakage, screw backout, plate impingement, inadequate compression across the fracture site as well as higher costs . Recently, two-HCS fixation has been advocated after extensively biomechanical investigating on fixation methods. A biomechanical study of Acar et al. using three-dimensional finite element analysis in three different wrist positions declared that two-HCS fixation imparted less displacement of the fracture gap and lower rotation of the fracture fragments than one-HCS fixation . Mandaleson et al.  compared three fixation options in a scaphoid nonunion model and reported that both double HCS and plate fixation had significantly greater stability, stiffness, and energy absorption than one-HCS fixation. No significant differences could be observed between two-HCS fixation and scaphoid plate fixation in any of the biomechanical parameters. The theoretical advantage of this technique is to achieve multiplanar stability by maximizing interfragmentary compression across the fracture site while minimizing shear and rotational forces that may hinder primary bone healing .
Garcia et al.  presented 19 patients with scaphoid nonunion using two-HCS fixation, showing superior rotational stability and a union rate of 100% without screw penetration and postoperative complications. Quadlbauer et al.  included a total of 47 patients with unstable scaphoid B2 type fracture and demonstrated that patients treated with two HCS had a significantly higher union rate than those with one HCS. EK et al.  showed good clinical outcomes and high union rates (90.5%) with two-HCS fixation and autologous distal radius cancellous bone grafting for delayed scaphoid unions and nonunions in a series of 21 patients. A retrospective study of 42 patients with scaphoid nonunions fixed with one versus two HCS found that two-HCS fixation had a higher union rate . With regard to screw length, Dodds et al.  suggested that the long screw was superior to the short one when checked for fracture fragment stability. We used two HCS measuring 18–26 mm for securing the fractures. We believe that the rigid fixation allows for early rehabilitation. Thus, all patients were advised to remove spica plaster and encouraged to initiate active ROM in the early stage after surgery.
Minimally invasive techniques are primarily indicated for minimally or nondisplaced scaphoid waist fractures and proximal pole fractures. A displacement greater than 1.0 mm is an indication of open reduction to obtain the anatomic alignment. However, these techniques have been successfully applied to the displaced scaphoid fractures, which usually require significant manipulation to correct the alignment. When this closed reduction cannot be achieved, open reduction is required. But the probability of such cases is relatively low . Moreover, minimally invasive techniques are not applicable to all types of scaphoid nonunions. They are indicated for early scaphoid nonunions without substantial bone absorption and loss of height with humpback deformity, and without AVN of the proximal pole. The chronic scaphoid nonunions with established structural deformities should be treated with more complicated procedures, usually involving open reduction, bone grafting and internal fixation . Although these techniques are successful, a thorough preoperative discussion is essential. Both the surgeons and patients must be aware of the risks of inadequate reduction and further redisplacement .
Slade et al.  first described the use of one percutaneous screw fixation under arthroscopy for the treatment of scaphoid nonunion. They reported good results with a union rate of 100% and a comparable ROM and grip strength of 87% compared to the contralateral hand at 1-year follow-up. Waitayawinyu et al.  studied a consecutive cohort of 22 patients treated with one HCS and an olecranon bone graft under arthroscopy, and all patients had satisfactory outcomes and bone healing. Wong et al.  found an overall union rate of 90.7% after arthroscopic treatment with one HCS and an iliac crest bone graft. The arthroscopic technique with two-HCS fixation and optional bone grafting has not been discussed to date. We recognize that the arthroscopic technique has several advantages. First, arthroscopy helps to visualize the fracture site and assess the fracture features. Adequate freshening of the fracture site could be easily performed, and punctate bleeding is also clearly visible. Second, concomitant injuries could be diagnosed under arthroscopic examination and treated at the same time. Third, external joystick reduction in the proximal and distal fracture fragments could be facilitated by the use of an arthroscopic probe. Moreover, it helps to visualize directly whether there is screw penetration above the articular surface of the scaphoid. It also contributes to minimizing surgical trauma to ligamentous structure and blood supply, less dissection of the soft tissue and the capsule, allowing for early restoration of wrist function. Additionally, it requires smaller skin incisions.
The gold standard for treating scaphoid nonunion is autogenous bone grafting. A randomized control trial reported similar results regarding union rate, time to union, and functional outcomes in 80 patients who received either vascular or nonvascular bone grafts, but three patients involved graft failure in the vascular bone graft group . The vascular bone grafting demanding microvascular procedures was complex, which required longer operation time, extensive dissection, limited pedicle length, and difficulty in obtaining and placing the graft [41, 42]. Nonvascularized bone grafts from the iliac crest and distal radius are most commonly used in scaphoid nonunion and proved to provide osteoconductive and other progenitor cells for accelerating bone healing . Garg et al.  conducted a prospective randomized trial of 100 patients to compare union rate and functional outcome between the two graft types. No significant difference was observed with a union rate of 87% for both. Tambe et al.  investigated 68 symptomatic scaphoid nonunion patients; the two graft types provided an equivalent union rate and radiologic outcomes, but a donor-site pain was found in 21% of patients undergoing iliac crest bone grafting. In a large systematic review by Pinder et al.  involving 1602 patients, they revealed that iliac crest bone graft had a comparable union rate with distal radius bone graft (87% and 89%, respectively), but a higher incidence of complications. We advocate the use of distal radius bone grafts as they are minimally invasive and require only regional anesthesia while iliac crest bone grafting usually requires general anesthesia. Furthermore, harvesting bone graft from the distal radius may increase the vascularity of the area and generate a form of regional bone repair response, as a result, improving the vascular flow in the wrist. This theoretical framework was proposed by Rellan et al. , where core decompression of the distal radius using an antegrade screw was proven to be successful in treating scaphoid nonunion. Our technique harvested adequate graft material from the distal radius to fill in the fracture site. Union was achieved in all 23 cases at an average of 11.6 weeks.
Besides the nonunion location and fixation stability, time to surgery was perceived as another major risk factor that resulted in the failure of treatment. Mahmoud et al.  reported that the average healing time was 9 weeks if time to surgery was less than 1 year, whereas an average healing time of 12 weeks was needed to achieve union if time to surgery was greater than 1 year. A multivariable regression analysis found a significant correlation between union rate and time to surgery in 160 scaphoid nonunion cases (a union rate of 92% if the nonunion duration was less than 5 years versus 80% if the nonunion duration was greater than 5 years) . It seemed plausible that progressively greater failure rates were associated with longer delays. Our study included patients with a wide range of time to surgery (from 1 day to 84 months). However, all patients achieved bony union with different healing time, which may prove that our technique is a valid treatment option for unstable scaphoid fracture and nonunion.
Several limitations of our study should be considered prior to interpretation. There was no control group including other techniques such as one HCS for a comparison with the experimental group. It is a retrospective follow-up investigation with a nonrandomized patient population, which may lead to selection bias. The small number of patients and the short-term clinical and radiographic follow-ups also limit the strength of our findings. We recognize that further high-quality studies with larger sample sizes and long-term follow-ups are warranted to determine the effectiveness and safety of this technique.
In conclusion, the arthroscopic technique with two-HCS fixation and distal radius bone grafting is a reliable and effective technique for the treatment of unstable scaphoid fracture and nonunion, providing satisfactory union rates and clinical outcomes.