Correlation between anthropometric measurements and graft size in anterior cruciate ligament reconstruction: a systematic review and meta-analysis

Purpose This systematic review and meta-analysis aimed to investigate the correlation between anthropometric measurements and graft size in anterior cruciate ligament (ACL) reconstruction. Methods A systematic search of Ovid MEDLINE, Embase, and Cochrane Library databases was conducted for observational studies published until March 2023 that reported the relationship between anthropometric data [height, weight, body mass index (BMI), age, gender, thigh length, and circumference] and ACL graft size. Correlation coefficients (COR) and their associated 95% confidence intervals were used as the primary effect size. This review was conducted in line with PRISMA guidelines. Results A total of 42 observational studies involving 7110 patients were included, with a mean age of 29.8 years. Statistically significant, moderately positive correlations were found between graft size and height (COR: 0.49; 95% CI: 0.41–0.57; p-value: < 0.001), weight (COR: 0.38; 95% CI: 0.31–0.44; p-value: < 0.001), thigh circumference (COR: 0.40; 95% CI: 0.19–0.58; p-value: < 0.001), and thigh length (COR: 0.35; 95% CI: 0.18–0.50; p-value: < 0.001). However, age and gender were insignificantly correlated with graft size (p-value: NS). A subanalysis based on graft type showed a significant positive correlation between height and graft diameter, which was more significant in the peroneus tendon than in hamstring grafts (COR: 0.76 vs. 0.45; p-value: 0.020). Conclusion This study demonstrated a moderate positive correlation between anthropometric measurements (height, weight, thigh circumference, and length) and ACL graft size, along with a weak positive correlation with BMI. Age and gender showed no significant correlation. These findings support the predictability and selection of ACL graft size based on pre-operative patient anthropometric data. Level of evidence Level of Evidence: IV. PROSPERO registration number: CRD42023416044. Supplementary Information The online version contains supplementary material available at 10.1007/s00590-023-03712-w.


Introduction
Anterior cruciate ligament (ACL) injury is a common knee injury with an incidence of up to 78 per 100,000 personyears [1].Surgical treatment is often required to restore knee biomechanics and function.Several autograft options are available for ACL reconstruction, such as bone-patellar tendon-bone (BTB), hamstring tendon (HT), quadriceps tendon (QUAD) and peroneal tendon (PLT) [2,3], while the popularity of hamstring tendon grafts has risen due to their biomechanical stability, low donor-site morbidity and improved fixation methods [4,5]; however, the success of

Data extraction and items
Two independent reviewers used a pre-designed data collection sheet in Microsoft Excel to extract data.The extracted demographic data included the first authors' surnames, study year, design and country, number of participants and knees, population type (adult vs paediatrics), graft type, the mean age of patients, gender, mean height, weight, BMI, thigh length and circumference, level of activity, correlations reported for each variable, statistical tests, and conclusions.

Qualitative assessment (risk of bias)
Two authors assessed the methodological quality of the included studies using the Methodological Index for Non-Randomized Studies (MINORS) assessment tool, which comprise eight key items, with a global ideal score of 16 for non-comparative studies [15].A higher overall score indicates a lower risk of bias; a score of 8 or less corresponds to a high risk of bias.

Statistical analysis
A meta-analysis of the eligible studies using R (version 4.0.2,R Core Team, Vienna, Austria, 2020) was conducted using the meta package (i.e.forest_meta and metacor).Correlation coefficients (COR) and their associated 95% confidence intervals were presented as the main effect size.For studies that reported beta regression values instead of Pearson's r, the latter was estimated using the equation r = 0.98ß + 0.5λ published by Peterson and Brown [16].Strength of the resultant effect sizes was interpreted per the criteria set by Cohen (x < 0.1, weak; 0.3 < x < 0.5, moderate; x > 0.5, strong) [17].Heterogeneity among effect sizes was evaluated using the I-squared statistic.Definitions for heterogeneity were adapted from the Cochrane handbook (< 25%, mild; 25-50%, moderate; > 50%, severe).Due to the high heterogeneity for the dichotomous variables, a randomeffects model was utilised.Both a funnel plot and Egger's test of asymmetry were utilised to assess publication bias.

Quality assessment [risk of bias and level of evidence (LoE)]
Based on the OCEBM criteria [18], 21 studies were level 2, 15 were level 3, and 6 were level 4 (Table 1), with an overall grade B of recommendation assigned to the review [19].The MINORS criteria scores of all 42 observational studies ranged from 10 to 15, with an average of 12.71 ± 1.29 (Out of 16), indicating a low overall risk of bias.A summary of the qualitative assessment, according to the MINORS criteria, is shown in the Supplementary material.

Subgroup analysis per graft type and region
When stratified by graft type, the correlation between age and graft diameter did not significantly differ between hamstring-and PLT-using studies (COR: 0.01 vs. 0.02; p-value: 0.580).Conversely, height was significantly more strongly correlated with graft diameter within PLT-using studies than their hamstring counterparts (COR: 0.76 vs. 0.45; p-value: 0.020).PLT-using studies demonstrated a strong correlation between weight and graft diameter compared to their hamstring-using counterparts; however, such difference was insignificant (COR: 0.64 vs. 0.35; p-value: 0.09).Similarly, differences in BMI correlation with graft diameter were statistically insignificant between PLT-and hamstring-using studies (COR: 0.32 vs. 0.15; p-value: 0.140).Stratification of correlations between anthropomorphic measures and graft diameter across different nations and graft types is provided in Tables 2 and 3.

Heterogeneity and publication bias
Significant heterogeneity was present across all pooled correlations ranging from 32.0 to 94.0%.Egger's test indicated funnel plot asymmetry for only the studies reporting on correlation between height and graft diameter (p = 0.004).Funnel plots for all pooled correlations are included within the supplementary material (Table 4).

Correlations between graft diameter and anthropomorphic measures
The correlation between age and graft diameter was deemed statistically insignificant.Clinically, this would be supported by evaluating the patient demographic undergoing ACLR.This would generally include the active adult population, in which muscular conditioning, development, and thus graft size would generally be considered comparable [20,21].
Where this correlation may be clinically significant would be in the elderly population, where ACLR may not be so readily performed due to poor-quality graft availability as a result of age-related sarcopenia [22,23].The weak insignificant correlation favouring an association between male gender and graft size should be treated with caution within the context of this review.This is partly due to the significant male predominance of the patients included in this review.Similarly, the literature on ACLR is still predominantly related to the male gender; however, this is shifting rapidly, and the considerations of female ACLR should be considered high on the agenda for future research priorities in soft tissue knee surgery [24][25][26].
Height, weight, thigh length and circumference all demonstrated a moderately positive correlation with graft size within this review.Such anthropomorphic measurements can be considered surrogate markers for muscular development, both in relation to cross-sectional area and axial muscular length and thus can be considered more relevant markers to base potential graft size upon.On the other hand, BMI demonstrated a weak correlation with graft size, supporting the notion that lean body mass calculation should be used in favour of BMI when considering eventual graft size, as reported in studies by Abatsi et al. [22,27].

Graft subgroup analysis
PLT-using studies demonstrated a strong correlation with height, weight, and graft diameter in comparison to hamstring-using studies.The reasons for this have not been born out in the literature but may support the notion that utilising the PLT as a graft of choice may have more reproducible and reliable clinical results if the treating clinician relies on anthropomorphic measurements in the pre-operative phase.However, to further validate these clinical conclusions, standardised methods of graft sizing and reporting would be required, and heterogeneity in their reporting within the context of this study may discredit any conclusions that can be drawn relating to the utility of different graft types.

Limitations
Anthropometric data should be used contextually, with generalisability not applicable between differing populations.For example, specific data relating to graft thickness in Caucasian populations may not correlate with recommendations for patients in South East Asia due to genetic differences in musculoskeletal structure between different populations [28].This review included data from various populations with subanalysis performed based on various regions; however, the skew was towards the Indian and American populations.Further work should generalise the analysis with equal representations from different populations.
This review predominantly focused on ACLR in the male population, with 94.8% of included patients male.Within ACLR, female patients experience high rates of graft-tunnel mismatch, laxity and re-rupture than male patients [29].This furthers the notion that future research  As surgeons gain more confidence in selecting appropriate graft types and planning surgeries based on anthropometric measurements, it could lead to better surgical outcomes.This, in turn, could contribute to reduced reoperation rates and healthcare costs, which may have implications for public health resource allocation.Also, improved pre-operative planning and graft size selection could potentially lead to fewer post-operative complications and revisions.This could alleviate the burden on the healthcare system, allowing resources to be directed towards other pressing health issues.

Fig. 1
Fig. 1 PRISMA flow diagram of record identification, screening and selection in metaanalysis

Fig. 2 Heterogeneity: I 2 =
Fig. 2 Forest plot of the overall pooled correlation between age and graft diameter.COR Correlation, CI confidence interval

Fig. 3
Fig. 3 Forest plot of the overall pooled correlation between gender and graft diameter.COR Correlation, CI confidence interval

Fig. 5
Fig. 5 Forest plot of the overall pooled correlation between weight and graft diameter.COR Correlation, CI confidence interval

Fig. 6
Fig. 6 Forest plot of the overall pooled correlation between BMI and graft diameter.COR Correlation, CI confidence interval

3
Subanalysis based on graft type comparing the correlation of anthropometric measures and graft size in Hamstring (HT) versus peroneus longus tendon (PLT) grafts BMI

Table 1
A summary of baseline study characteristics

Table 1
LoE level of evidence, FU (Y) follow-up in years, HT hamstring, PLT peroneus longus tendon, BMI body mass index

Table 2 A
summary of the primary correlation analysis of anthropometric measures and ACL graft size

Table 4
Subanalysis based on region comparing the correlation of anthropometric measures and graft size in Hamstring (HT) versus peroneus longus tendon (PLT) grafts