Cement augmentation for proximal humerus fractures: a meta-analysis of randomized trials and observational studies

A comprehensive search in electronic databases (PubMed, Embase, and CENTRAL) for studies on cement augmentation for proximal humerus fractures was performed. Table S1 in the S1 File describes the full search syntax. The search was performed on December 1, 2023. All randomized controlled trials and observational studies that compared cement augmentation with no augmentation in plate fixation of proximal humerus fractures were included in this review. Other inclusion criteria were reporting on the outcomes of interest and availability of full text.

Exclusion criteria were cadaveric studies, studies on pathologic fractures, case reports, and studies with languages other than English, Dutch, French, German, Spanish, or Italian.

Two reviewers assessed the search and inclusion of studies independently (YL, BW). Disagreement was solved by consensus with a third reviewer (FB).

Study and patient characteristics were collected in a predefined data extraction sheet and included first author, publication year, country/region in which the study was performed, study design, study population size, operating position/approach, and type of cement used. Furthermore, fracture type according to the Neer classification, gender, and the patient’s history of smoking or diabetes were extracted [11].

The same two reviewers (YL, BW) assessed the methodological quality of included studies, independently using the MINORS criteria [12]. Disagreement was resolved by consensus. Details are described in Table S2 in the S1 File.

The primary outcome was overall complication rate. Additionally, complications were subdivided in fracture/implant-related and systemic. Fracture/implant-related complications included screw perforation/cut-out, total or partial humeral head necrosis, non-union, loss of reduction, implant bending or breakage, and deep wound infection. Non-union was defined as the absence of callus formation or fading of fracture lines on X-ray after 6 months. Loss of reduction was ≥ 15° increase of varus malposition and a relative change of ≥ 5 mm of the greater or lesser tuberosity [13].

Systemic complications included thromboembolic complications, pneumonia, and renal insufficiency. Also, a separate analysis specifically addressing the risk of humeral head necrosis and screw perforation was performed.

Secondary outcomes included re-interventions, hospital stay, operation time, range of motion, functional scores, and general quality of life measured at 12 months after the operation. Re-interventions included all re-operations performed on the affected fracture site during follow-up. Functional scores included Constant score and DASH score [14, 15]. General quality of life was measured using the EQ5D score [16].

Continuous variables were presented as means with standard deviation (SD) or range. If required, information was converted to mean and SD using the methods described in the Cochrane Handbook for Systematic Reviews of Interventions. Dichotomous variables were presented as counts and percentages.

Formal meta-analysis was only performed in case three or more studies reported on the outcome of interest. Effects of treatment options on continuous outcomes were pooled using the (random effects) inverse variance weighting method. They were presented as mean difference (MD) with corresponding 95% confidence interval (95%CI). Binary outcomes were analyzed using the (random effects) Mantel–Haenszel method. They were presented as odds ratio (OR) or risk difference (RD), with a 95%CI.

Heterogeneity between studies was quantified by the I² statistic. The threshold for significance was set at a p-value of 0.05. Review Manager (RevMan, version 5.4.1) was used for statistical analysis.

A total of 651 references were evaluated. A detailed description of the search and screening is shown in Table S1 in the S1 File. Five observational studies and one randomized controlled trial fulfilled the criteria [13, 17‐21].

The six studies included a total number of 541 patients, of which 249 received cement augmentation following fixation. Baseline characteristics including age, gender, Neer classification, operative approach, and type of cement are described in Table 1. They were equally distributed among treatment groups. All studies used a deltopectoral approach. Polyaxial locking-compression plates were used in all studies.

The mean quality of all studies was 19 points (range 16–22) using the MINORS criteria. Details can be found in Table S3 in the S1 File.

Overall complications were reported in five studies: one randomized clinical trial and four observational studies. The overall complication rate was significantly lower in the augmented group (15.6% versus 25.4%, RD 11%, 95%CI 4–18%) with an OR of 0.54 (95%CI 0.33–0.87, I² 0%) (Fig. 1).

The difference in overall complications between treatment groups was predominantly caused by a difference in implant-related complications, which occurred in 10.4% in the augmented versus 19.9% in the non-cemented group with an OR of 0.49 (95%CI 0.28, 0.88, I² 18%). All implant-related complications are described in Table 2. The most common complication was secondary screw protrusion, which occurred more frequently in the non-cemented group (1.5% vs. 12.2%, OR 0.18, 95%CI 0.05–0.64, I² 0%) (Fig. 2). Loss of fixation was not significantly different between the groups (3.3% in the augmented group vs. 7.7% in the non-augmented group, OR 0.43, 95% CI 0.13, 1.39, I² 0%). There was no significant difference in humeral head necrosis (5.6% in the augmented vs. 4.1% in the non-augmented group, OR 1.39, 95%CI 0.64, 3.04, I² 0%).

To date, this is the first formal meta-analysis on this topic. A systematic review on the use of cement augmentation in proximal humerus fractures including clinical and biomechanical studies was published in 2020 [5]. The authors pointed out the possible benefits of the technique in terms of stability (i.e., citing the reduction in screw penetration from 16.6 to 8% in the study of Katthagen et al. or the loss of fixation in 10.9% (cemented) vs. 5.1% (non-cemented) reported by Siebenbuerger et al.). Due to the limited clinical data, however, no clear recommendation could be made. Additionally, concerns regarding the safety of the procedure, mainly the risk of cement-related humeral head necrosis, were pointed out. With two new large-scale studies published in the past 2 years, almost double the number of patients were available creating the opportunity to investigate the benefit of cement on a meta-analytical level. With this dataset, the possible benefits and the safety of the procedure in terms of cement-related complications could be investigated more thoroughly.

Based on the results of this meta-analysis, one could argue that cement augmentation should be routinely used in elderly patients treated with plate fixation for proximal humerus fractures. However, certain aspects should be considered.

The benefit of cement augmentation appears to lie predominantly in reducing the risk of implant failure, more specifically secondary screw protrusion. Interestingly, one would expect to find a lower re-intervention rate in the cemented group. This, however, is not the case. A logical explanation might lie in the study population itself. Elderly patients with proximal humerus fractures generally have an advanced age and low demands. The threshold to perform a second operation in this fragile patient group is high. These characteristics combined with the fact that implant failures do not necessarily cause severe complaints might reduce the need and desire for revision surgery. On the other hand, the fact that no difference in re-intervention was found might very well be caused by underpowering as only three studies reported on this outcome.

Regarding adverse effects, a major concern in cement augmentation is humeral head necrosis. This meta-analysis showed no significant difference regarding this complication. All studies, however, did point in the same direction indicating that it might occur more often in the cemented group. Nevertheless, the absolute risks in this meta-analysis were low (5.6% in the augmented vs. 4.1% in the non-augmented group). Indeed, much higher risks up to 35% have been described in previous literature [22]. This is however mainly attributable to the fact that their study population contained more four-part fractures, a well-known risk factor for humeral head necrosis, than included in the present meta-analysis. In other words, it is plausible to assume that humeral head necrosis is mostly dictated by fracture morphology rather than whether cement augmentation was used or not.

Regarding the functional results, no difference was found with little heterogeneity. This seems a logical finding as cement mainly has a biomechanical advantage. It reinforces the construct and does not directly improve functional outcomes such as range of motion. Also, functional results were measured at approximately 1 year. It is a well-known fact that measuring functional scores at 1 year follow-up carries the risk of measuring coping instead of true function. The best possible time to measure functional scores is 6 months. This data, regrettably, was not available in this meta-analysis.

It should be questioned whether the additional costs of cement augmentation including the costs for fenestrated/cannulated screws are justified. Indeed, it reduces the risk of implant failure; however, it remains uncertain whether this has any clinical consequences. Material costs of cement augmentation are approximately 500$ [23]. Costs related to a potential increased operation duration seem negligible based on the results of this meta-analysis.

To our opinion, cement augmentation is justified, but whether to use it should be determined on a case-by-case basis. It should be underlined that cement augmentation cannot be used as a salvage procedure in poorly reduced fractures. Good patient selection and anatomical reduction of the fracture remain the most decisive factors in preventing complications. To our opinion, patients that will benefit the most from it are those with a fracture pattern with acceptable risks for humeral head necrosis where anatomic reduction is achieved but the osteosynthesis requires additional anchorage due to poor bone quality or little bony substance surrounding the screw tips.

In this meta-analysis, observational studies were mainly included. Only one randomized clinical trial was available. Although previous meta-analyses have shown that pooled analysis of observational studies demonstrates similar risk estimates as the ones of randomized clinical trials in orthopedic trauma research, this assumption could not be internally validated in the present meta-analysis. Furthermore, it should be noted that heterogeneity in some of the analyses was quite high, i.e., for re-intervention and operative time, making these results less reliable.

Furthermore, the quality of the studies included varied considerably. Many also lacked clear definitions on study outcomes such as humeral head necrosis and wound infection.