Introduction

Facial asymmetry can signify conditions such as hemifacial microsomia (HFM), a congenital craniofacial disorder attributable to uneven development of the face stemming from abnormalities in the first and second pharyngeal arches [1]. HFM, with an occurrence rate of approximately 1 in every 5500–26,000 live births, is ranked as the second most prevalent congenital craniofacial abnormality [2]. It affects multiple facial components, including the skeleton, ears, nerves, and soft tissues [3]. In contrast with isolated microtia, which is characterized solely by ear abnormalities, HFM typically exhibits pronounced facial asymmetry. This asymmetry can significantly impact patients' physical and mental well-being, underscoring the critical need for enhanced research in this domain [4].

There is a lack of comprehensive, quantitative studies featuring robust design and precise assessments in the research on facial asymmetry in HFM [5]. The debate also continues over whether isolated microtia represents the mildest form of HFM, as research has not yet thoroughly explored the facial geometry of isolated microtia or its distinction from HFM [6, 7]. Furthermore, there is no consensus on the progression of facial asymmetry in HFM, an understanding crucial for effective treatment planning. Research findings remain inconclusive, attributed to factors such as limited study designs, small sample sizes, methodological issues, and occasionally questionable conclusions.

This cross-sectional study is designed to comprehensively analyze facial asymmetry in cases of HFM and isolated microtia. It will assess asymmetry in both the skeletal structures and soft tissues across various facial regions in two age groups, utilizing precise three-dimensional imaging. Additionally, it intends to compare the facial asymmetry in these conditions and examine the evolvement of facial asymmetry.

Materials and Methods

This cross-sectional study was granted ethical approval by the Ethics Committee of Peking Union Medical College. It focused on the analysis of pre-treatment computed tomography (CT) images obtained from East Asian patients at the Center for Auricular Reconstruction, Plastic Surgery Hospital, which is affiliated with the Chinese Academy of Medical Sciences, spanning the period from 2020 to 2023. Pre-treatment CT images were acquired for diagnostic and therapeutic purposes, adhering to the manufacturer's guidelines for the CT scanner (CT 6000, Philips Healthcare, Netherlands), with imaging parameters including a 1-mm slice thickness, 120 kV, and 75 mA.

A total of 120 male patients aged 5–12 years with unilateral HFM or isolated microtia were included (see Table 1). Inclusion criteria are as follows: confirmed diagnosis of HFM or isolated microtia unilaterally, mild mandibular deformities (Pruzansky–Kaban types I and IIA) in HFM, the availability of pre-treatment CT images, absence of other craniofacial syndromes, and no history of maxillomandibular surgery or trauma. The diagnosis of the patients was carried out collaboratively by two experienced plastic surgeons.

Table 1 Demographics for all cases and separate groups
Full size table

Based on characteristics of craniofacial development and prior research [8,9,10], patients who met the inclusion criteria were categorized into four groups by age and diagnosis: (1) HFM group, ages 5–7 (n = 30); (2) HFM group, ages 8–12 (n = 30); (3) isolated microtia group, ages 5–7 (n = 30); and (4) isolated microtia group, ages 8–12 (n = 30).

Extraction and Registration of Hard and Soft Tissue

The DICOM images were imported into Mimics software (Materialise, Leuven, Belgium) for segmentation. Threshold values for hard tissues were established at 226–3071 Hounsfield units (HU) and for soft tissues at − 250–3071 HU. The segmented images were then imported into 3-Matic software (Materialise, Leuven, Belgium) for further processing.

According to the previous studies, thirteen landmarks both hard or soft tissues on the unaffected side and the middle were manually placed. Then based on the previous studies, proper midsagittal plane for patients with craniofacial deformity was defined [4]. Later, we delineated other planes and then segmented the facial area into eight subunits and a whole, using plane 8, the coronal plane, 7 and 9 to define the facial borders upside down [11,12,13,14] (see Tables 2 and 3, Fig. 1a–c). Eight anatomical units were subsequently delineated based on the previous studies of facial esthetic subunits [15] and facial fat compartments [16] (see Figs. 1d and 2).

Table 2 Description of facial landmarks on hard and soft tissues
Full size table
Table 3 Description of reference planes
Full size table
Fig. 1
figure 1

Comprehensive facial mapping: landmarks and segmentation techniques for hard and soft tissues. a (above left) Seven facial landmarks on hard tissue; b (above right) six facial landmarks on soft tissue; c (below left) twelve reference planes; the blue, brown, and green dotted lines indicate the median, horizontal, and coronal planes, respectively; the red dotted lines indicate other reference planes for segmentation; and d (below right) eight facial units segmented according to facial esthetic subunits on both tissues

Full size image
Fig. 2
figure 2

Segmentation process of anatomical regions. After removing the ears on both sides and defining all planes, the entire head undergoes segmentation of facial borders and regions

Full size image

Using the midsagittal plane as a reference, the anatomical regions were mirrored to ensure alignment of the left and right structures. Subsequently, the software conducted an unsigned point-based part comparison analysis to evaluate surface deviation between overlapped anatomical regions and to generate a heat map image (see Fig. 3). Root-mean-square (RMS) values for surface deviation across all nine units, which include eight anatomical units and the entire face, were subsequently collected.

Fig. 3
figure 3

The process for conducting symmetric analysis of craniofacial tissues. The original face is mirrored using median plane as a reference and then overlapped with the mirror face; a heat map showing the surface deviation is then generated

Full size image

Statistical Analysis

Statistical analysis utilized SPSS v20 (Chicago, III) software. The Shapiro–Wilk test indicated that the distributions of age and RMS values for surface deviation in both hard and soft tissues were non-normal (see Table 4). Differences between two independent groups were assessed using the Mann–Whitney U-test with two-tailed tests employed. Effect sizes from the Mann–Whitney tests were categorized as small (r ≤ 0.1), small to medium (0.1 < r < 0.3), medium (r = 0.3), medium to large (0.3 < r < 0.5), and large (r ≥ 0.5). Effect sizes were deemed clinically significant when medium or larger. The correlation between age and RMS values in merged groups of patients with the same disease was assessed using the Spearman correlation coefficient. A significance level of P < 0.05 was established.

Table 4 The Shapiro–Wilk test results of RMS values for surface deviation of hard and soft tissues in various facial regions, grouped by condition and age
Full size table

Test of Reproducibility of the Procedure

For reproducibility assessment, 10 patients with HFM and 10 with isolated microtia were randomly selected from the total cohort of 120 subjects. The anatomical regions were then re-extracted twice, 14 days apart, by the same operator. Subsequently, the RMS values were remeasured, and intra-class coefficients were calculated, showing that the extraction and segmentation of the hard and soft tissue units were reproducible and responsible (see Table 5).

Table 5 The intraclass coefficients results of RMS values for surface deviation of hard and soft tissues in various facial regions
Full size table

Results

Condition Groups Comparisons

Differences in median RMS values (mm) between the HFM and isolated microtia groups for ages 5–7 ranged from 0.01 to 1.1, with the 8–12 age groups from 0.06 to 0.98 (Table 6). Significant differences were observed in most regions, with the HFM groups displaying larger RMS values overall, except in the forehead and nasal units of soft tissue, where differences were not statistically significant (Table 7).

Table 6 Descriptive statistics of RMS values for surface deviation of hard and soft tissues in various facial regions, grouped by condition and age
Full size table
Table 7 Mann–Whitney U-test results comparing RMS values for surface deviation differences of hard and soft tissues in various facial regions between groups
Full size table

Age Groups Comparisons

For both HFM groups, median RMS values in hard and soft tissues were observed to share similar ranges, from 0.8 to 2.3 and 0.9 to 2.3, with differences remaining consistently below 0.3 across all areas. The greatest asymmetry was displayed by the lower cheek unit, with median RMS reaching up to 2.0 in hard and 2.3 in soft tissues (Table 6). With the exception of the nasal and medium cheek areas in hard tissues, where the 8–12 HFM group exhibited larger RMS values, most regions demonstrated no significant differences between groups (Table 7). Median RMS values for the 5–7 and 8–12 isolated microtia groups ranged from 0.6 to 1.4 in both tissues, with intergroup differences remaining below 0.2. Notably, the upper-mid cheek units in hard and soft tissues showed the relative high asymmetry (Table 6). Significant age group differences were only identified in the nasal and upper lip areas of hard tissues for isolated microtia patients and in the nasal and median cheek area of hard tissues for patients with HFM, suggesting an increase in asymmetry with age (Table 7).

Age–Facial Asymmetry Correlation

Table 8 shows the Spearman correlation coefficients, indicating a moderate association (P < 0.05, 0.25 < r < 0.50) between age and RMS values in specific facial regions. For the HFM group, age showed a moderate correlation with the medium cheek and nasal units. In the isolated microtia group, a similar correlation was found with the nasal and upper lip units. Other facial areas did not exhibit significant associations.

Table 8 Results of the Spearman correlation coefficient between age and RMS values for surface deviation of hard and soft tissues in various facial regions, grouped by condition
Full size table

Discussion

Disputed Relationship Between HFM and Isolated Microtia

This study provides insights into facial asymmetry, utilizing one of the largest cohorts of patients with HFM and isolated microtia examined to date, and quantifies for the first time the differences between these two conditions, offering a more comprehensive understanding of their facial geometry. Both hard and soft tissue asymmetries were examined, with analyses conducted on regional and whole face asymmetry simultaneously. Significant discrepancies were identified, particularly in the mid-lower facial regions. These results were expected given that facial dysmorphology selectively affects maxillary and mandibular structures of patients with HFM [17]. Additionally, based on the knowledge that the forehead and nasal units both belong to the upper-third of the trigeminal area selection which corresponds to different embryological origins [18]. However, it was unexpected that the asymmetry differences of forehead and nasal units in hard and soft tissues showed inconsistency, with differences only significant in hard tissue. This may be attributed to the soft tissue's masking effect on hard tissue asymmetry [19, 20]. Besides, as our findings indicate that HFM exhibits the least asymmetry in the forehead and nasal units, these results might suggest that the deformity of HFM extends to the upper facial area, and unfortunately, soft tissue in the mid-lower face fails to fully mask the hard tissue asymmetry. Besides the significant discrepancies, the regions exhibiting the highest and lowest RMS values, along with those showing minimal RMS differences across age groups, also vary in two conditions. Therefore, it is imperative to consider specific asymmetry distribution patterns in diagnosis and treatment management.

The classification of isolated microtia as the mildest form of HFM continues to be debated [21]. Proponents of this classification cite similarities in ear malformations, higher incidence rates of microtia, and related anomalies in families with HFM as their rationale [22]. However, the recently introduced European HFM Guidelines do not recognize microtia as a mild HFM form [21, 23]. The quantitative findings from our study generally support separate examinations of the two conditions from a new dimension. Additionally, an overlap in the RMS ranges of these conditions was observed, suggesting that not all patients with HFM exhibit noticeable facial asymmetry, highlighting the challenge in management protocols making and distinguishing between isolated microtia and some HFM cases with sole skeletal defects.

Disputed Progression of Facial Asymmetry

The large cohort and well-controlled samples in our study offer greater statistical power compared to many previous reports. Our study concentrates on facial asymmetry during childhood and preadolescence, a period characterized by consistently high growth velocities prior to the onset of adolescence. We discovered that while asymmetry in certain hard tissue regions may vary over time, it generally remains stable across most facial areas, encompassing both hard and soft tissues. Furthermore, we noted consistent patterns of asymmetry within the same condition across different age groups, as well as comparable levels of asymmetry across various conditions within the same age group, as detailed in Table 7.

Our findings align with the previous research on the stability of facial asymmetry in the normal pediatric population [24]. The observed correlation between age and asymmetry in midline regions may be attributed to differential growth phases of facial bones and the robust connectivity within the midfacial skeletal structures, including the nasal and maxillary skeletons [25, 26]. However, the literature presents conflicting views regarding the progression of facial asymmetry in patients with hemifacial microsomia (HFM). Kearns et al. analyzed facial angles relative to age and noted variations that suggest changes over time [7]. Similar findings were reported in follow-up studies by Kaban et al. and Ara et al. [27, 28]. Conversely, Meazzini et al. observed that the ratios between affected and unaffected ramal heights in HFM patients remained consistent throughout growth when left untreated [29]. These discrepancies in research outcomes likely stem from differences in sample selection, methodological approaches, and inherent limitations across studies. Challenges such as limited follow-up, selection bias, and inadequate control of confounding factors such as age, gender, and disease type are prevalent [30]. The use of small sample sizes and traditional 2D imaging techniques, which often result in distorted views, further complicates the accurate assessment of facial asymmetry [31, 32]. These constraints highlight the necessity for more robust and comprehensive evaluation methods in our study. However, due to the absence of a universally accepted method for assessing craniofacial asymmetry, direct comparison of results may not be feasible.

Our study introduces advancements in the measurement techniques and ensures homogeneity of the sample in terms of age and severity of the pathology. The results reveal that compared to patients with isolated microtia, those with hemifacial microsomia (HFM) exhibit an age-related increase in hard tissue asymmetry within the median region, which includes both mandibular and maxillary components. This suggests a progressive deterioration of the deformity in HFM patients over time. Despite this progression in hard tissue deformity, we observe that the asymmetry of soft tissues across all regions remains stable. Notably, this includes asymmetries in bone, muscle, and fat, indicating a compensatory effect from the soft tissues.

Diagnosis and Management

Our study emphasizes the importance of understanding the relationship between HFM and isolated microtia for defining, diagnosing, and facial asymmetry management in these conditions. Furthermore, the presence of subtle facial asymmetry in patients with HFM prompts concerns regarding underdiagnosis in those with solely skeletal deformities and emphasizes the risk of excessive use of radiographic examinations in patients with isolated microtia not in need of skeletal corrective treatment.

In terms of treatment, surgeons may opt to correct facial deformities in patients either during early development or after maturity, with treatment preferences varying among surgeons at different therapy stages [33]. For patients with HFM classified as Pruzansky–Kaban types I and IIA, treatment options include unilateral functional appliance treatment, fixed orthodontic treatment, distraction osteogenesis, and facial fat grafting [34, 35]. Our study observes that facial bone asymmetry, particularly in the median cheek region of patients with mild hemifacial microsomia (HFM), increases over time. Given the progressive nature of the deformity, we advocate for early surgical intervention in children with HFM to support optimal growth and prevent further secondary distortions. However, several studies suggest that such treatments do not significantly alter the natural trajectory of facial growth, indicating that early orthopedic interventions may not substantially reduce asymmetry [36, 37]. Consequently, we recommend a cautious approach to early, single-stage interventions, especially in mild cases.

To focus our research, we studied two age groups of East Asian males with HFM limited to Pruzansky–Kaban types I and IIA, recognizing that significant craniofacial growth occurs mainly during childhood and adolescence, and HFM primarily affects males. This approach, aimed at reducing confounding factors, may limit the broad applicability of our findings. Instead of following strict anatomical guidelines, we segmented facial hard tissues into esthetic subunits to better understand the relationship between skeletal structures and soft tissues. This method, while not precise for individual bone analysis, offers insights into esthetic impacts, aligning with the needs of cosmetic and reconstructive surgery. Although using a semi-automatic method for image segmentation and labeling could introduce subjective errors, we ensured high image quality and utilized established craniofacial landmarks for accurate and reproducible measurements.

Conclusion

In conclusion, our study provides crucial reference data on the asymmetry of both soft and hard facial tissues in patients with HFM and isolated microtia as they grow. We highlight the importance of recognizing the distinct differences in facial asymmetry between these two conditions, as well as their evolvement over time, in their diagnosis and treatment. Although the specific characteristics of these disorders remain to be fully elucidated, our methods and results offer valuable insights for further research into these unique conditions.