Locking plate fixation for proximal humerus fractures: are four proximal screws adequate? A clinical and radiological analysis of 105 consecutive cases

Abstract

Introduction

Proximal humerus fractures are common injuries, and achieving stable fixation is essential to prevent complications such as varus collapse and screw perforation. While optimal screw positioning has been emphasized biomechanically, clinical evidence regarding the required number of screws in the humeral head, particularly when minimally invasive aiming devices limit placement to four screws, remains limited. This study evaluated whether fixation with four cranial screws provides adequate clinical and radiological stability.

Materials and methods

A retrospective cohort study was conducted including 105 patients with displaced proximal humerus fractures treated with plate osteosynthesis. Patients were grouped according to the number of proximal screws used: four screws (n = 58) or more than four screws (n = 47). The primary outcome was the Constant–Murley Score. Secondary outcomes included secondary screw perforation, head–shaft angle maintenance, restoration of medial cortical support, and avascular necrosis. Mean follow-up was 18 ± 6 months.

Results

Functional outcomes were comparable between groups, with Constant–Murley Scores of 78 ± 7 in the four-screw group and 73 ± 13 in the multi-screw group (p = 0.23). Rates of screw perforation (8.6% vs. 10.6%; p = 0.72) and postoperative head–shaft angles were likewise similar, and no significant alignment change occurred during follow-up. Medial cortical support was restored more frequently in the multi-screw group (82% vs. 63%; p = 0.001), but this did not translate into superior clinical or radiographic outcomes. Avascular necrosis occurred in 3 and 13% of patients, respectively.

Conclusion

When fracture reduction/alignment is adequately restored, the use of four cranial screws appears sufficient to maintain alignment and prevent mechanical complications in minimally invasive plate osteosynthesis of proximal humerus fractures. Further prospective studies with longer follow-up are required to validate these findings.

Introduction

Proximal humerus fractures are among the most frequently observed injuries, and their incidence is increasing in the elderly population [1–4]. For grossly displaced fracture patterns, surgical stabilization using plate or nail osteosynthesis is generally considered the preferred treatment modality [5–8]. Nevertheless, surgical therapy remains associated with considerable complication rates, with implant loosening, loss of reduction, and screw cut-out being among the most common adverse events [9].

Achieving anatomic reduction and preventing complications are essential prerequisites for favorable functional outcomes [10]. Several factors have been identified to influence the risk of complications. In addition to patient-related parameters such as age and bone mineral density (BMD), obtaining anatomic reduction and restoration of medial cortical support are critical determinants [11]. Biomechanical studies have further emphasized the relevance of optimal screw placement to gain stable fixation [12–14]. However, existing studies have primarily focused on the biomechanical aspects of the medial cortical support [15–17]. Clinical data regarding the required number of screws in the humeral head remain limited. This knowledge gap is particularly relevant because percutaneous aiming devices used for locking plate osteosynthesis often permit the placement of only four screws in the humeral head [18].

The purpose of this study was therefore to determine whether the four most cranial screws placed in the humeral head provide sufficient fracture fixation in a clinical setting.

Materials and methods

Study design

A retrospective cohort study was conducted at a level 1 trauma center. All patients presenting with a traumatic proximal humerus fracture classified as AO/OTA type 11-A: C who underwent minimally invasive plate osteosynthesis (MIPO) between 2007 and 2010 were considered eligible for inclusion. Exclusion criteria comprised patients younger than 18 years, individuals with pre-existing conditions such as hemiplegia or other relevant neurological disorders, those with polytrauma defined by an Injury Severity Score (ISS) greater than 15, an open reduction technique, fracture dislocation type, and patients with post-traumatic brachial plexus injury or peripheral nerve palsy.

This study was approved by the local ethics committee (KEK-ZH-Nr. 2010 − 0421/4).

Data collection

Patients were identified using our prospective trauma database. Baseline characteristics (e.g. age, sex, mechanism of injury, fracture classification, and any associated injuries) were obtained from this source. Data regarding operative management (such as surgical duration, approach, number of screws placed in the humeral head, and the use of a calcar screw), as well as postoperative complications, were extracted from the electronic hospital records.

Fractures were classified according to the original AO fracture classification system published in 1990 [19], as all included cases were treated prior to the introduction of the revised 2018 AO/OTA classification. Radiological assessment included evaluation of the initial fracture displacement (varus or valgus) and measurement of the head–shaft angle. This angle was determined on the 20° external rotation antero-posterior radiograph by comparing a tangent to the articular surface with a line parallel to the long axis of the humeral shaft, as described previously [20] (Fig. 1).

Fig. 1.

Head-shaft angle measured on the 20° external rotation AP radiograph between the articular tangent and humeral shaft axis [20]

Additional information for the assessment of mid- to long-term functional outcomes after operative stabilisation was obtained during routine outpatient follow-up. At each follow-up visit, the Constant Murley score of both the injured and uninjured shoulder was recorded [21]. The active range of motion was measured using a goniometer, and strength testing was performed with an Isobex isometric digital dynamometer (MDS, Oberburg, Switzerland). Clinical evaluation also included assessment for impairment of the anterior branch of the axillary nerve Figs. 2 and 3.

Fig. 2.

Percutaneous aiming device for the PHILOS^®-System (DePuySynthes^®, Switzerland). Due to the required axillary nerve safe zone, only four screws can be inserted proximally and four distally when using the 5-hole plate

Fig. 3.

Percutaneous PHILOS^®-System (DePuySynthes^®, Switzerland) aiming device in situ with non-absorbable sutures passed through the rotator cuff tendons and corresponding plate holes

Group allocation

Two patient cohorts were defined based on the proximal screw configuration determined by the available instrumentation at the time of surgery. Before the introduction of the radiolucent aiming device for the PHILOS^® system (DePuySynthes^®, Switzerland), proximal fixation could be achieved with more than four screws through the minimally invasive approach. These patients constituted group two (multi-screw fixation). After the radiolucent aiming device became available, safe screw placement was limited to four proximal screws to avoid potential injury to the axillary nerve. Patients treated thereafter formed group one (four-screw fixation) (Fig. 4).

Fig. 4.

Annual distribution of proximal humerus fracture fixations using four versus more than four head screws, illustrating the shift toward the four-screw configuration following introduction of the aiming device

Indication for operative treatment

Indications for operative treatment were a varus displacement of > 20°, a valgus displacement of > 40°, an increased reclination > 30°, a lateral displacement of > ½ diaphyseal diameter, and/or displacement of the major and/or minor tubercle of > 5–10 mm [8].

Operative procedure

Patients were placed in the beach-chair position, which allowed optimal fluoroscopic visualisation in multiple planes using the image intensifier. In all cases, a five-hole PHILOS^®-Plate (DePuySynthes^®, Switzerland) allowing angle stable screw fixation in the humerus head was used. With the availability of the radiolucent aiming device for PHILOS^®-System (DePuySynthes^®, Switzerland), only four screws could be used in the humeral head, therefore reducing the number of screws for proximal fixation (Figs. 2 and 3). The indication for fixation, as described above, was the same in both groups. All cases were operated using a minimal invasive antero-lateral delta-split approach splitting the delta muscle at the level of the anterior raphe and avoiding extending the incision for more than 6 cm distal to the anterior tip of the acromion as described previously [18]. Non-absorbable sutures were placed through the rotator cuff tendons and knotted onto the plate for firm fixation and reduction of the tuberosity in all cases [18]. In the four-screw fixation group, the radiolucent aiming device required placement of the screws in the two most proximal rows of the five-hole PHILOS^®-Plate (DePuySynthes^®, Switzerland) only, according to rows A and B. In contrast, patients in the multi-screw fixation group received additional proximal screws in the remaining rows (C–E) as deemed necessary by the operating surgeon.

Postoperative treatment

Postoperative management was identical in both groups and consisted of immediate initiation of physiotherapy with restricted weight bearing. Use of a sling was recommended only for analgesic purposes during the first few days after discharge. On the second postoperative day, as well as at 6 weeks postoperatively, fracture reduction and implant positioning were verified using bidirectional X-rays. At the 6-week follow-up, depending on the radiographic evidence of fracture healing, a pain-adapted increase in weight bearing was permitted until full weight bearing was achieved.

Follow up

Routine clinical and radiological follow-up examinations were scheduled at 6 and 12 weeks, as well as at 6 and 12 months postoperatively in the outpatient department. Additional follow-up visits were arranged upon patient request or when clinically indicated.

Outcome measures

The primary outcome was the Constant - Murley Score [21]. This clinician-based assessment evaluates overall shoulder function on a scale from 0 (severe dysfunction) to 100 (normal function), incorporating pain, activities of daily living, range of motion, and strength. A difference of approximately 10 points is generally considered clinically meaningful.

Secondary outcome parameters were secondary screw perforation, the head–shaft angle, restoration of medial cortical support, and the occurrence of avascular necrosis (AVN). Additional secondary outcomes included revision surgery (e.g., correction of postoperative malreduction or screw exchange) and other complications, such as nonunion [22], malunion [22], infection [23], implant failure, and any procedure requiring correction of the initial osteosynthesis.

Statistical analysis

Patient data are reported as n (%) or n ± SD and outcome data are reported as mean. The matched numerical outcome was measured using the Wilcoxon test and one-way ANOVA for functional outcome in multiple groups. A p-value < 0.05 was considered significant. We used a minimal important clinical difference (MICD) of the Constant score of 6.7 [24]. To achieve a power of 0.8 and a significance level of 0.5 with a MICD of 6.7 score difference, 42 samples were necessary in each group. SPSS^® statistics version 22 (IBM, Chicago, Illinois, USA) was used for the statistical analyses.

Results

During the study period, 105 patients with displaced proximal humerus fractures treated with minimally invasive plate osteosynthesis met the inclusion criteria. In total, a proximal fixation with four screws was used in 58 patients (group one), whereas more than four screws were inserted in 47 patients (group two). The mean follow-up for all patients was 18 ± 6 months (minimum 12 months).

Baseline characteristics are presented in Table 1. The mean patient age was 61 ± 13 years. The dominant arm was affected in 42 patients (40%). According to the AO/OTA classification, 55% of all fractures were type B. Ski accidents (39%), pedestrian injuries (26%), and domestic falls (18%) accounted for more than 80% of the mechanisms of injury. Most patients were classified as ASA I or II (> 90% in both groups). Seventeen board-certified trauma or orthopedic surgeons performed the procedures.

Table 1.

Baseline characteristics

Groups	4 screws	>4 screws	p value
Patients	58	47
Gender
Female	34	29	0.8
Age (y ± SD)	59 ± 14y	62 ± 12	0.55
Dominant (n)
Yes	19	23	0.4
Accident (n)
Ski	24	17	0.58
Fall at home	9	10	0.44
Pedestrian	14	13	0.86
AO Classification (n)
A	13	8	0.49
B	36	22	0.11
C	9	17	0.015

SD standard deviation

Primary outcome

The mean Constant - Murley Score was 78± 7 in group one and 73 ± 13 in group two, with no significant difference between groups (p = 0.23) (Table 2).

Table 2.

Primary and secondary outcome parameters after proximal humerus fracture fixation with four or more proximal screws

Groups	4 screws	> 4 screws	p value
Follow-up (months ± SD)	16.8 ± 5.9	20.2 ± 6.5	0.004
Postoperative head-shaft angle (degree ± SD) Restored medial cortical support	42° ± 7° 63%	40° ± 8° 82%	0.06 0.001
OR Time (min ± SD)	69 ± 21	92 ± 30	< 0.05
Screw perforation	8.6%	10.6%	0.72
Constant Score (± SD)	78 ± 7	73 ± 13	0.23

SD standard deviation

Secondary outcomes

Secondary screw perforation occurred in 8.6% of patients in group one and 10.6% in group two, with no significant difference (p = 0.72). Postoperative head–shaft angles were comparable between groups, measuring 42° ± 7 in group one and 40° ± 8 in group two (p = 0.06) (Table 2). At the final available follow-up, no significant change from the postoperative alignment was observed in either group (p = 0.97) (Fig. 5). Medial cortical support was restored significantly more often in group two (39 fractures, 82%) compared with group one (37 fractures, 63%; p = 0.001). Avascular necrosis of the humeral head was documented in 2 cases (3%) in group one and 6 cases (13%) in group two.

Fig. 5.

Head–shaft angle distribution postoperatively and at 18 ± 6 months in patients treated with either four or more than four cranial screws

Discussion

In this study, proximal fixation with only four screws demonstrated favorable clinical and radiological outcomes after a mean follow-up of approximately 1.5 years. Functional results were comparable between groups, with no significant difference in the Constant–Murley Score, suggesting that limiting the number of proximal screws does not compromise shoulder function. Likewise, we found no relevant difference in secondary varus or valgus displacement or in rates of screw perforation between treatment groups. These findings support that, when combined with adequate fracture reduction, a reduced number of proximal screws may provide sufficient stability in minimally invasive plate osteosynthesis.

Restoration of the physiological head–shaft angle remains a critical component of successful proximal humerus fracture management, as both biomechanical stability and physiological shoulder kinematics depend on it. Previous work has highlighted the negative effects of varus malalignment, demonstrating that increasing varus angulation leads to higher muscular demand and impaired kinematics, with deviations exceeding 20° considered unacceptable [10, 11]. In our cohort, restoration of the overall head–shaft alignment was achieved in both groups, with postoperative head–shaft angles within the expected physiological range and without significant changes during follow-up. These radiographic findings align with the low rates of secondary screw perforation and support the biomechanical integrity of both fixation strategies.

The optimal number and configuration of screws in the humeral head required to achieve stable fixation and prevent varus collapse remain subjects of ongoing debate. Biomechanical studies addressing this issue have reported heterogeneous findings. Kim et al. showed in a biomechanical two-part proximal humerus fracture model without inferomedial support that increasing the number of proximal screws improved construct stiffness and load to failure, with the 7-screw group showing the most favorable mechanical properties. However, differences compared with the other screw configurations did not reach statistical significance, and the authors acknowledged that only isolated varus shear loading was investigated, which incompletely reflects the multidirectional loading conditions encountered in vivo [14]. Bai et al. demonstrated a significant increase in axial and shear stiffness with the addition of a calcar screw in a simulated unstable two-part fracture model. However, no further biomechanical benefit was observed once medial cortical support had already been restored [17]. In contrast, Katthagen et al. found no significant improvement in construct stability adding an additional infero-medially directed calcar screw [25]. Moreover, Shin et al. reported that inferomedial supporting screws did not show additional biomechanical advantages in non-osteoporotic proximal humerus fractures beyond increasing overall construct stiffness. The specific contribution of inferomedial screws to fracture stability remained unclear, although a greater number of screws and the use of inferomedial screws appeared more resistant to cyclic loading [16]. Our clinical findings align with these biomechanical observations. Despite a significantly higher rate of restored medial cortical support in the group with more than four proximal screws, this did not translate into superior functional outcomes, improved maintenance of the head–shaft angle, or reduced rates of secondary screw perforation.

From an anatomical and biomechanical perspective, the rationale for emphasizing cranial screw placement is further supported by studies investigating bone quality and load distribution within the humeral head. Hepp et al. demonstrated that BMD is highest in the cranial region of the humeral head, suggesting superior screw purchase in this area [26]. These findings were subsequently confirmed by Schiuma et al., who likewise identified the cranial aspect as the region of greatest osseous density [27]. These anatomical considerations are complemented by biomechanical data on load distribution, indicating that the most cranial screws bear approximately 60% of the load transmitted through the construct, whereas the most caudal screws contribute only about 10% [28]. Therefore, the effectiveness of inferior screw placement appears limited, given the comparatively lower bone quality in this region and its reduced contribution to load transmission. In the present study, all proximal screws were intentionally placed as long as possible to achieve subcortical support in the humeral head, where bony resistance is maximal. Despite the use of only four proximal screws, our findings suggest that screw quality and strategic placement within regions of optimal bone stock are more critical than the absolute number of screws used.

Conversely, restoration of medial cortical support has consistently been identified as a critical predictor of fixation failure [11]. Numerous studies have confirmed the importance of the medial cortex as a key stabilizing structure in proximal humerus fracture fixation [12, 13, 29, 30]. Direct medial support is typically achieved through anatomic reduction of the two main fracture fragments - the humeral head and shaft. However, in the presence of severe osteoporosis and/or additional medial metaphyseal comminution, direct cortical contact may not be achievable. In such cases, we intentionally fix the fracture with a slight shortening (approximately 5 to a maximum of 10 mm) combined with controlled valgisation by manual axial compression prior to insertion of the diaphyseal screws. When medial comminution precludes direct head–shaft contact, the medial zone is bridged in correct length using a non-touch, minimally invasive technique. We believe that this approach preserves the vascularity of the medial fracture fragments and promotes early callus formation and fracture consolidation. Interestingly, in the present cohort, restoration of medial cortical support was significantly less frequent in the group treated with only four proximal screws. Despite this, no increase in varus collapse or inferior clinical or radiological outcomes was observed.

Limiting proximal fixation to four screws inherently results in a longer bridging segment across the fracture site, which may promote indirect metaphyseal fracture healing by allowing controlled micromotion. In contrast to very stiff constructs created by occupying nearly all available plate holes, a more flexible fixation may facilitate callus formation and potentially reduce the risk of secondary screw cut-out. This concept is consistent with our clinical findings, as stable alignment and low rates of secondary screw perforation were observed despite the reduced number of proximal screws.

Several limitations of this study must be acknowledged. First, its retrospective design inherently limits the ability to control for all confounding variables. Second, as a single center analysis, selection bias cannot be excluded. Our institution treats a substantial number of sport-related injuries, resulting in a relatively young and healthy patient population. Although BMD was not measured, it is possible that patients in our cohort had higher BMD than those treated in other settings, potentially facilitating improved screw purchase in the humeral head and influencing the generalisability of our findings. Third, fracture severity was not evenly distributed over time, with significantly more AO/OTA type C fractures in the multi-screw cohort. Fracture morphology may therefore represent an important confounding factor independent of screw number. In addition, the transition to the four-screw technique coincided with the introduction of the radiolucent aiming device for the PHILOS^®-Plate system. We acknowledge that subgroup analyses were limited by the relatively small cohort size and the retrospective study design. Consequently, the independent effects of screw number, fracture complexity, and surgical experience cannot be fully separated in this retrospective analysis. Conversely, the high level of standardisation of the surgical technique at our centre, combined with possible effects of a surgeon learning curve, may have influenced treatment decisions and outcomes. Fourth, although the mean follow-up duration was 18 months, the minimum follow-up was 12 months. Loss to long-term follow-up is a known challenge in orthopaedic trauma research, particularly in cohorts where many patients become asymptomatic and decline further evaluation [31]. While our follow-up duration was sufficient to assess fracture healing, implant stability, and early complications, it limits the ability to fully determine the incidence of avascular necrosis, which may become clinically or radiographically evident after two or more years after the index surgery. Therefore, the observed rates of osteonecrosis may underestimate the true incidence.

Conclusion

Our findings suggest that, when adequate fracture reduction/alignment is achieved, a fixation strategy using four cranial screws may provide sufficient stability to prevent varus collapse and secondary screw cut-out in minimally invasive plate osteosynthesis of proximal humerus fractures. Prospective studies with larger patient cohorts and longer follow-up periods are needed to confirm these findings.

Author contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by C.M., C.S. and YP.A. PF.S and C.Z. made substantial contributions and revised the work critically for important intellectual content. The first draft of the manuscript was written by C.M. and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

No funding was received. Each author certifies that he has no commercial associations that might pose a conflict of interest in connection with the submitted article.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Conflict of interest

The authors declare no competing interests.