Influence of premolar extractions on long-term adult facial aesthetics and apparent age

Summary

Objectives:

This study aimed to assess whether non-extraction, two-premolars and four-premolars extraction Class II treatment protocols are different regarding long-term facial aesthetics, age appearance, and soft-tissue measures.

Methods:

Frontal and lateral photographs of 63 full Class II division 1 patients, treated at least 8 years before, were evaluated by 83 laypeople and 76 orthodontists, who assigned to each one’s attractiveness scores from 1 to 10, and opined about their apparent ages. Patients were divided in three groups: XP0 (non-extraction, n = 20; 30.77 years, 15.63 years post-treatment), XP2 (two-premolars extractions, n = 25; 30.99 years, 15.68 years post-treatment), and XP4 (four-premolar extractions, n = 18; 32.80 years, 18.01 years post-treatment), that were matched by gender, post-treatment occlusal and soft-tissue outcomes, age, and post-treatment time. Soft-tissue measures were obtained with Dolphin Imaging 11.5 software.

Results:

Two-way analysis of variance revealed that attractiveness and apparent age were not affected by treatment protocols, but laypeople were slightly more critical then orthodontists. Analysis of variance showed more vertical facial pattern in XP4 group. Pearson correlation test revealed no influence of soft-tissue measures on sample’s attractiveness.

Conclusions:

Treatment of full Class II division 1 malocclusion with and without extractions did not influence facial attractiveness, age appearance, and overall soft-tissue measures in the long-term.

Introduction

Nowadays facial aesthetics is strongly valued, sometimes even more than occlusion and functions, and consists in one of the primary goals of orthodontic treatment. Fuller lips and slightly protrusive profile are being preferred, especially in women (1, 2). Several studies concluded that smaller noses may also be desired (3–7). Retruded lips may emphasize the size of nose and chin, making them look bigger (6, 8).

Concerns with the ‘dished in’ profile has prevented some orthodontists from choosing treatment plans based on premolar extractions (9). Speculation that extractions may harm facial aesthetics is based on the reduction of dental volume available for lip support. Even many years after they have been introduced in the orthodontic scenario, extraction effects are still being studied regarding efficiency, occlusal outcomes, stability, and post-treatment aesthetics (10–15).

In Orthodontics, premolar extraction may be indicated for severe crowding, excess dentoalveolar protrusion and skeletal/inter-arch discrepancies camouflage (14). Several authors have evaluated the effects of extractions on patient’s faces (14, 16–20). However, most of them only compare cephalometric results obtained with and without extractions, not considering the actual judgment of facial aesthetics (16–20), neither the long-term effects (16–20). More important than cephalometric final values are the overall facial attractiveness provided by treatment.

After the end of growth, there are still continuous changes in the nose and chin, increasing lips retrusion (21). Because protruded lips are related to attractive faces (2), relative increases of other structures negatively influence profile attractiveness. As these soft tissue changes somehow contribute to an expected decrease in attractiveness, concerns with the soft tissue effects of premolar removal may increase with time. But what happens in the long-term? Is post-retention facial aesthetics the same in patients treated with and without premolar extractions?

Class II malocclusion treatment may involve two-maxillary premolar extractions generally when there is no crowding or cephalometric discrepancy in the mandibular arch, or two maxillary and two mandibular premolar extractions, primarily to correct crowding in the mandibular arch, a cephalometric discrepancy, or a combination of both, in growing patients (22). To our knowledge, no study has evaluated the long-term influence of premolar extractions on Class II division 1 malocclusions, exclusively. This context led us to assess if non-extraction, two-premolar extraction and four-premolar extraction treatment protocols of Class II division 1 malocclusion are different regarding facial aesthetics and the apparent age in the long-term.

Material and methods

This study was approved by the Research Ethical Committee of Bauru Dental School, University of São Paulo (number 032/2011). Informed consents were obtained from patients and raters.

Sample

Patient sample size calculation was based on an alpha significance level of 0.05 and a beta of 0.2, to achieve a power of 80%. To detect a difference of 1.25mm in the distance from the upper lip to H line with an estimated standard deviation of 1.3mm (23), each group should have at least 18 patients. The number of raters were similarly calculated to detect a mean difference of 0.5 point in the rating scores, with an estimated standard deviation of 1.06 point (24), and was found to be of 72 raters.

Sixty-three orthodontically treated phenotypically mediterranean patients were selected according to the following inclusion criteria: 1. initial full cusp Class II division 1 malocclusion, 2. adult—minimum of 21 years of age (25); 3. availability of orthodontic records; 4. no anterior tooth loss in the long-term period; and 5. frontal and lateral extra-oral photographs, taken at a minimum of 8 years after treatment. Patients were divided into three groups according to the treatment protocol: non-extraction (n = 20), two-maxillary premolar extractions (n = 25) and four-premolar extractions (n = 18).

Orthodontic mechanics included fixed edgewise appliances, with 0.022-inch × 0.028-inch conventional brackets and an usual wire sequence characterized by initial 0.015 Twist-Flex or 0.016 Nitinol, followed by 0.016-, 0.018-, 0.020-, and 0.019-inch × 0.025-inch or 0.018-inch × 0.025-inch stainless steel archwires (all from 3M Unitek, Monrovia, California, USA). Deepbites were corrected with accentuated and reversed curve of Spee (26). In the extraction groups the anterior teeth were retracted en masse with rectangular archwires for overjet and Class II canine correction.

Extraoral headgear was used to correct the Class II antero-posterior relationship in the non-extraction and in the four-premolar extraction groups; in the non-extraction group, six patients used functional appliances and three patients used them associated to extraoral headgear. In the two-premolar extraction group, extraoral headgear was used to reinforce anchorage and maintain the Class II molar relationship. When necessary, Class II elastics were used to help obtain a Class I molar relationship, in the non-extraction and in the four-premolar extraction groups, and to help maintain a Class II molar relationship in the two-premolar extraction group. A maxillary Hawley plate was recommended to be used full-time (except during meals) for 6 months and at night, for additional 6 months. A mandibular canine-to-canine fixed retainer was installed to be used for a mean period of 3 years.

Groups comparability

The three groups had to present similar occlusal and cephalometric orthodontic treatment outcomes. The post-treatment occlusal outcomes were evaluated using the objective grading system (OGS) index of the American Board of Orthodontics (27, 28) and specific soft tissue variables. The cephalometric comparability was evaluated with the post-treatment cephalograms.

The post-treatment lateral headfilms, obtained in centric occlusion with passive lip posture, were scanned and stored on a personal computer and analysed with Dolphin Imaging 11.5 Software (Dolphin Imaging and Management Solutions, Chatsworth, California, USA) (29). Because the headfilms had been taken with different X-ray machines, the enlargement factors, which ranged from 6.0% to 9.8%, were corrected with the cephalometric software. The variables obtained on the final lateral cephalograms are displayed in Table 1. To check intergroup comparability regarding occlusal stability, the long-term overjets were clinically measured and well matched.

Table 1.

Soft-tissue measures obtained from the long-term lateral photographs and from the post-treatment cephalograms.

Field	Variable	Landmarks and lines	Abbreviation	Obtained on
Soft-tissue measures	Frontonasal angle (°)	G’N’Prn	FNA	LTP
	Nasal proeminence (°)	PrnN’Sn	NP	LTP
	Nasal length (mm)	Sn-Prn	NL	LTP
	Tip of nose—H line (mm)	Prn-Pg’Ls	H-nose	BOTH
	Nasolabial angle (°)	ColSnLs	NLA	BOTH
	Mentolabial angle (°)	LiB’Pg’	MLA	LTP
	Mentocervical line (mm)	C’-Me’	MCL	LTP
	Cervicofacial angle (°)	SnPg’Me’C’	CFA	LTP
	Facial angle (°)	SnPg’G’	FA	LTP
Distance from lips to aesthetic lines	Lower lip-H line (mm)	Li-Pg’Sn	Li-H	BOTH
	Upper lip-E line (mm)	Ls-Pg’Prn	Ls-E	BOTH
	Lower lip-E line (mm)	Li-Pg’Prn	Li-E	BOTH
	Upper lip-S line (mm)	Ls-Pg’Cm	Ls-S	BOTH
	Lower lip-S line(mm)	Li-Pg’Cm	Li-S	BOTH
	Upper lip-Pg’Sn line (mm)	Ls-Pg’Sn	Ls-Pg’Sn	BOTH
	Lower lip-Pg’Sn line (mm)	Li-Pg’Sn	Li-Pg’Sn	BOTH
	Upper lip-Pg’N’ line (mm)	Ls-Pg’N	Ls-Pg’N’	LTP
	Lower lip-Pg’N’ line (mm)	Li-Pg’N’	Li-Pg’N’	LTP
Growth pattern	Upper lip length (mm)	Sn-Stms	ULL	LTP
	Lower lip length (mm)	Sn-Stmi	LLL	LTP
	Lower anterior face height (mm)	Sn-Me’	LAFH	LTP
Dental measure	Post-treatment overjet	Dolphin defined	—	PTC
	Long-term overjet		—	CLN

LTP, obtained on the long-term lateral photographs; PTC, obtained on the post-treatment cephalograms; BOTH, obtained on both images, in both stages; CLN, clinically measured.

Clinical photographs

Long-term post-treatment lateral and frontal extra-oral photographs were obtained in a standardized position. Subjects were oriented to stay in natural head position (NHP)—a reproducible, upright, and natural posture, with the visual axis horizontally oriented, teeth in centric occlusion, lips at rest. They were asked to not to touch the wall with their back, to keep arms relaxed and feet slightly apart, looking at their eyes height (30). Any accessories like earrings, necklaces, or piercings were removed. Long hairs were homogeneously tied. All photographs were taken with a digital Nikon D90 SLR camera (Nikon Corporation, Tokyo, Japan) coupled to a Sigma 105mm 2.8 DG MACRO lens and a Sigma EM-140 DG NA-iTTL ring flash (Sigma Corporation, Kanagawa, Japan).

These photographs were used for the aesthetic and apparent age judgment. The lateral photographs were also used to obtain cephalometric soft tissue variables (31). The soft-tissue variables were obtained with Dolphin Imaging Software, which are also listed on Table 1. A known distance of each subject was inserted into the software to calibrate each photograph to its actual size. The planes and lines used are illustrated in Figure 1.

Figure 1.

Planes, lines, and angles involved on the soft-tissue measures obtained from the long-term lateral photographs.

Aesthetic and apparent age judgment

Using Adobe Photoshop CS6 (Adobe Systems Incorporated, version 13.0, San Jose, CA, USA), the Frankfurt plane was checked to be horizontally oriented on the lateral images and the midsagittal plane was checked to be vertically oriented in the frontal images. Any distractors such as spots, pimples, scars, or tattoos were removed. After that, all images were converted to grey scale (Figure 2). The frontal and lateral photographs of each patient were combined to consist in a pair of images. Therefore, considering that each patient had long-term post-treatment stage lateral and frontal photographs, there was a total of 126 images, distributed in 63 pairs of images.

Figure 2.

Converting original photograph on the final image for aesthetic judgment.

Based on a previous methodology (24), potential raters were invited by e-mail to access the website created for this research, with provided personal usernames and passwords. Raters registered date of birth, gender, area of formal education and higher reached level. In order to calibrate themselves regarding the sample, raters were instructed to first evaluate the Gallery section, where all 63 patients could be simultaneously observed (Figure 3). The pairs of photographs were automatically randomized, in each access. There was no time restriction for each evaluation, until the final evaluation was submitted.

Figure 3.

Section Gallery, where the raters could visualize all 63 subjects to calibrate themselves regarding the sample’s attractiveness.

The raters used a 10-point numerical scale, where 1 represented ‘the most unattractive face’ and 10 represented ‘the most attractive face’. They were also asked to evaluate each patient’s apparent age (Figure 4). The difference between each apparent and actual age was then calculated. When the apparent was greater than the actual age, the difference between them was positive, which meant that the patient appeared to be older than he/she actually was and vice-versa. An icon in each page could drive the rater to the Gallery section to revise the scores, as necessary, allowing intra-rater calibration in evaluating the faces.

Figure 4.

The 10-point numerical scale under each pair of photographs and the question about each individual’s apparent age.

Raters

The aesthetic evaluation was carried out by 76 orthodontists from a national Orthodontic Association (39 post-graduate orthodontic students and 39 specialists) and a panel of 83 lay people (4 physicians, 7 from other biomedical areas, 18 professionals from exact sciences, 39 professionals from human sciences, and 15 described themselves as not fitting in none of the options).

Error study

Two weeks after the first evaluation, 23 randomly selected photographs were retraced by the same examiner (Cintia Helena Zingaretti Junqueira). Twenty-five orthodontists and 28 laypeople re-evaluated the same subjects in a different arrangement. Random errors were calculated according to Dahlberg’s formula (S ² = Σd ²/2n) (32), where S ² is the error variance and d is the difference between two determinations of the same variable. Systematic errors were estimated with dependent t-tests, for P < 0.05 (33).

Statistical analyses

Normal distribution of all data was verified by Shapiro–Wilk tests. From the variables obtained on cephalograms, almost all had normal distribution, except, post-treatment and long-term overjets. Among the variables obtained on the long-term lateral photographs, upper and lower lip lengths and lower anterior face height did not show normal distribution. Therefore, all these variables were compared using non-parametric versions of statistic tests.

Intergroup gender distribution was evaluated with chi-square test. Intergroup comparability regarding occlusal outcomes (OGS), soft-tissue profile, long-term age and post-treatment time were evaluated by analysis of variance (ANOVA). Intergroup comparability regarding post-treatment and long-term post-treatment overjet was compared by Kruskal–Wallis test.

Inter-rater group gender distribution and age were evaluated with chi-square and Mann–Whitney tests, respectively.

The influence of the treatment protocol, rater group, and of their interaction on facial aesthetics were evaluated with two-way ANOVA. The same was used for the influence of the apparent age, rater group, and of their interaction on facial aesthetics.

The long-term post-treatment photographic soft-tissue variables between the different groups were compared with ANOVA and Kruskal–Wallis test, followed by Tukey tests. Correlations between soft-tissue variables and facial attractiveness were evaluated with Pearson’s correlation coefficient. Results were regarded as significant at P < 0.05. These analyses were performed with Statistica software (Version 7.0; StatSoft Inc., Tulsa, Oklahoma, USA).

Results

Regarding aesthetic judgment, only one statistically significant systematic error was found, in the apparent age judged by laypeople, and there were no random errors. Regarding the soft-tissue variables, systematic errors were found in CML, Li-S, and Li-Pg’Sn. Random errors were found in NLA = 1.97, LMA = 2.73, CML = 1.98, LLL = 1.37, and AIFH = 1.68.

The groups were comparable regarding all variables, except the nasolabial angle at the post-treatment stage (Table 2).

Table 2.

Comparability between groups regarding gender; post-treatment soft-tissue profile, orthodontic outcomes and overjet; long-term overjet, age and post-treatment time.

	Groups	XP0 (N = 20)		XP2 (N = 25)		XP4 (N = 18)		P value
		N	%	N	%	N	%
	Male	9	25.71	13	37.14	13	37.14
		Mean	SD	Mean	SD	Mean	SD
Post-treatment stage	Orthodontic outcomes (OGS)	42.30A	16.9	41.68A	9.53	46.17A	12.89	0.516₳
	NLA (mm)	111.52A	11.13	117.39B	8.65	110.09A	10.68	0.045₳*
	H-nose (mm)	6.19A	5.46	5.48A	4.29	4.93A	5.22	0.733₳
	Li-H (mm)	0.08A	1.93	0.84A	1.33	1.13A	1.77	0.136₳
	Ls-E (mm)	−3.50A	3.17	−3.13A	2.40	−2.82A	3.03	0.761₳
	Li-E (mm)	−2.09A	3.41	−1.06A	2.59	−0.59A	3.00	0.285₳
	Ls-S (mm)	−0.49A	2.82	−0.42A	2.12	0.17A	2.79	0.685₳
	Li-S (mm)	−0.29A	3.12	0.57A	2.43	1.18A	2.80	0.265₳
	Ls-Pg’Sn (mm)	3.38A	2.53	3.06A	1.89	3.88A	2.54	0.522₳
	Li-Pg’Sn₳ (mm)	2.06A	2.81	2.66A	2.23	3.41A	2.56	0.263₳
	Overjet (mm)	2.88A	1.47	2.49A	0.8	2.84A	0.72	0.180ķ
Long-term stage	Overjet (mm)	3.52A	2.00	2.69A	0.93	3.22A	0.62	0.120ķ
	Age (years)	30.77A	7.14	30.99A	5.26	32.80A	5.77	0.530₳
	Period time (years)	15.63A	7.21	15.68A	4.83	18.01A	5.71	0.370₳

₢, Qui-square test; ₳, analysis of variance; ķ, Kruskal–Wallis test; NLA, nasolabial angle, Ls, upper lip; Li, lower lip.

*P < 0.05.

The rater groups were comparable regarding gender, but laypeople were significantly younger than orthodontists (Table 3).

Table 3.

Rater groups comparison regarding gender (Qui-square test) and age (Mann–Whitney test).

Variable	Laypeople		Orthodontists		P value
Gender
Female	55	56.12%	43	43.88%	0.210
Male	28	45.90%	33	54.10%
Age	35.52 (SD 11.18)		39.12 (SD 9.12)		0.003*

*P < 0.05.

Regarding facial aesthetics and apparent age, there were no significant intergroup differences (Tables 4 and 5). Laypeople attributed significantly smaller scores for facial aesthetics and greater apparent ages than orthodontists. The interaction between the groups and raters did not affect facial aesthetics or apparent age.

Table 4.

Comparison of aesthetic scores considering the influence of treatment protocols and rater groups (two-way analysis of variance).

Criteria																P value
Groups
	XP0 (N = 40)€					XP2 (N = 50)€					XP4 (N = 36)€					0.088
	Mean	SD				Mean	SD				Mean	SD
	4.73	0.81				4.73	0.91				4.22	0.64
Raters
		Laypeople (N = 83)							Orthodontists (N = 76)							0.000*
		Mean	SD						Mean		SD
		4.32	1.23						4.80		1.33
Groups × Raters
XP0 (N = 20)		XP0 (N = 20)		XP2 (N = 25)			XP2 (N = 25)			XP4 (N = 18)		XP4 (N = 18)				0.782
Laypeople		Orthodontists		Laypeople			Orthodontists			Laypeople		Orthodontists
Mean	SD	Mean	SD	Mean	SD		Mean	SD		Mean	SD	Mean	SD
4.51	0.76	4.97	0.88	4.47	0.89		5.00	0.96		4.00	0.69	4.46	0.66

€The number of observations is doubled because there are two groups of raters.

*P < 0.05.

Table 5.

Comparison of age appearance, considering the influence of treatment protocols and rater groups (two-way analysis of variance).

Criteria																			P value
Groups
	XP0 (N = 40)€									XP2 (N = 50)€					XP4 (N = 36)€				0.273
	Mean				SD					Mean	SD				Mean	SD
	2.27				4.96					-0.05	4.73				1.13	4.54
Raters
			Laypeople (N = 83)										Orthodontists (N = 76)						0.000*
					Mean		SD						Mean		SD
					1.34		4.86						0.68		4.72
Groups × Raters
XP0 (N = 20)				XP0 (N = 20)				XP2 (N = 25)			XP2 (N = 25)			XP4 (N = 18)		XP4(N = 18)			0.469
Laypeople				Orthodontists				Laypeople			Orthodontists			Laypeople		Orthodontists
Mean		SD		Mean		SD		Mean		SD	Mean	SD		Mean	SD	Mean		SD
2.61		4.98		1.89		4.98		0.31		4.80	-0.44	4.69		1.34	4.76	0.90		4.34

€The number of observations is doubled because there are two groups of raters.

*P < 0.05.

Upper and lower lip lengths were significantly longer in the four-premolar extraction group than in the other two groups and its lower anterior face height was greater than the non-extraction group. There was no individual significant correlation between any of the variables and facial attractiveness (Table 6).

Table 6.

Intergroup comparison regarding long-term soft-tissue variables and their correlation with the aesthetic scores.

	XP0 (N = 20)		XP2 (N = 25)		XP4 (N = 18)		P value	Pearson
	Mean	SD	Mean	SD	Mean	SD
Soft-tissue measures
FNA (°)	141.76A	5.1	142.54A	5.82	143.09A	7.86	0.804₳	0.532
NP (°)	22.11A	2.32	21.82A	2.74	20.82A	2.63	0.278₳	0.51
NL (mm)	19.77A	2.12	19.68A	2.23	19.02A	1.9	0.492₳	0.929
H-nose (mm)	10.43A	5.37	9.97A	4.94	9.86A	7.73	0.951₳	0.416
NLA (°)	108.68A	8.41	110.1A	10.22	103.84A	12.14	0.141₳	0.245
MLA (°)	128.3A	11.72	132.64A	13.93	134.13A	8.98	0.293₳	0.476
CML (°)	37.85A	13.87	38.8A	7.49	37.48A	7.35	0.902₳	0.923
CFA (°)	116.85A	14.22	118.52A	8.39	116.82A	9.57	0.834₳	0.106
FA (°)	13.59A	6.34	13.9A	5.92	11.53A	5.47	0.405₳	0.472
Distance from lips to aesthetic lines
Li-H (mm)	−0.6A	1.53	0.04A	1.44	0.59A	2.09	0.101₳	0.714
Ls-E (mm)	−5.78A	3.07	−5.68A	2.94	−5.63A	4.5	0.991₳	0.383
Li-E (mm)	−4.25A	2.72	−3.52A	3.09	−3.13A	4.71	0.607₳	0.342
Ls-S (mm)	−2.37A	2.6	−2.3A	2.35	−2.01A	4.08	0.929₳	0.48
Li-S (mm)	−2.13A	2.41	−1.43A	2.64	−0.84A	4.34	0.46₳	0.45
Ls-Pg’Sn (mm)	2.08A	2.17	2.07A	1.87	2.53A	3.63	0.821₳	0.648
Li-Pg’Sn (mm)	0.68A	2.06	1.28A	2.14	2.02A	3.89	0.322₳	0.636
Ls-Pg’N’ (mm)	7.39A	3.49	7.79A	3.04	7.64A	4.77	0.937₳	0.675
Li-Pg’N’ (mm)	3.97A	2.83	4.86A	2.85	5.13A	4.15	0.511₳	0.713
Growth pattern
ULL (mm)	21.48A	1.74	22.00A	3.13	24.16B	3	0.006ķ*	0.622
LLL (mm)	46.15A	4.59	48.39A	6.66	50.96B	4.28	0.012ķ*	0.117
LAFH (mm)	68.16A	5.36	71.37AB	9.27	76.08B	6.23	0.005ķ*	0.36

₳, analysis of variance; ķ, Kruskal–Wallis test; FNA, frontonasal angle; NP, nasal proeminence; NL, nasal length; NLA, nasolabial angle; MLA, mentolabial angle; CFA, cervicofacial angle; FA, facial angle; ULL, upper lip length; LLL, lower lip length, LAFH, lower anterior facial height; Ls, upper lip; Li, lower lip.

*P < 0.05.

Discussion

Sample selection

Different types of malocclusion in the same sample can provide misleading results. For this reason, only Class II division 1 malocclusions were included in this investigation. Besides presenting increased overjet and proclined maxillary incisors, it is also associated to increased ANB angle and more retrusive jaws (34, 35), standardizing the sample profile.

The influence of three Class II malocclusion treatment protocols on facial aesthetics were investigated in the long-term. It was assumed that the best treatment protocol was chosen for each patient at the time they were treated. It was not the aim of this study to investigate whether the treatment protocol applied to each patient was the best option.

The groups were well matched regarding the occlusal and cephalometric treatment outcomes, with only the two-maxillary premolar extraction group showing significantly larger nasolabial angle than the other two groups (Table 2). The influence of this difference will be addressed.

Concerns in conducting an actual long-term post-treatment evaluation reduced the chances for a larger number of participants because after a long time the patients phone numbers and addresses may have changed.

Raters selection

Facial aesthetics concepts can vary widely, because it is subjective and personal. To measure facial aesthetics, wide and different groups opinions should be considered (19, 36). Therefore, besides the 76 orthodontists from a national Orthodontic Association, 83 laypeople, from different professional areas were included.

Methodology

Facial attractiveness is affected by the way it is evaluated (37). Because orthodontic treatment usually intends to improve the facial profile, it is very often studied in the orthodontic literature. However, people are usually presented to society from a frontal view. Consequently, frontal and profile photographs were selected to be used in this study. Smiling photographs were excluded because any dental irregularity could impair the aesthetic evaluation.

Profile aesthetic studies sometimes focus on the lower third (23), or hide attributes as smile, lips, eyes, skin, under profile silhouettes (38–41) or contour lines (14). However, it is not the way profiles are evaluated in everyday life (42), which is as a part of a whole face. Although facial attributes are able to influence facial attractiveness judgment (43), they have larger impacts separately than when judged in a context (44). Raters preferences in photographs are closer to established aesthetic norms than in silhouettes (45,46). Consequently, the speculated aging effect of premolar extractions, if existent, should clearly appear. That would be shown by eventual lack of soft-tissue support, or deep sulcus in the anteroinferior facial third, such as the nasogenian sulcus.

To avoid an exhausting process, there was no time limit for all 63 evaluations. The 10-point numerical scale (24) provided a simple, non-suggestive method of evaluation, reporting precisely and quantitatively the raters opinions. Reliability was ensured because no effort was made to guide the evaluation with attractive or unattractive examples. The calibration Gallery section may have contributed to good reproducibility, calibrating the raters.

The only systematic error in the aesthetic evaluation suggests that nonprofessional aesthetic scores may vary depending on the moment. Coming from a lay group, and related to a quite subjective topic, this is an understandable result. Random errors were observed in angles in which their landmarks are very close, such as NLA and MLA, that are more likely to show statistically significant errors in their measurements (47).

Facial attractiveness

According to the raters, there was no statistically significant difference in facial attractiveness between the Class II treatment protocols, which agrees with some previous studies (23,48–50) (Table 4). These studies did not confirm that premolar removal would be harmful to the face. However, all of them emphasized that to be successful, premolar extractions should be carefully indicated and managed. The significantly greater initial nasolabial angle of group 2 did not influence the results (Table 2).

This similarity could be expected in the comparison between the non-extraction and the two-maxillary premolar extraction groups because the post-treatment soft-tissue profiles are usually similar in these protocols (51). Therefore, it is very reasonable that they remain similar in the long-term.

Laypeople were generally more critical than orthodontists, assigning significantly lower scores to the subjects attractiveness (Table 4), likewise previous results (17). These different views may be due to their different ages. Laypeople group, significantly younger than the professional group, was less tolerant to what they considered lack of attractiveness. Professional raters, slightly older and quite used to accurate facial analysis, probably valued individual attributes that the lay ones did not (52).

Apparent age

The decrease in dental volume due to premolar extractions may raise more concerns in the long-term (21,53–56). Assessing the subjects apparent ages was a way to verify a supposed aging effect of the extraction protocols on their faces.

Estimation of apparent age has been found in plastic surgery literature (57, 58). Groups of judges evaluated perceived ages analysing standardized frontal photographs of several patients, resulting in consistent results regarding the effects of surgical procedures and of the environment in age appearance (57, 58). Although the importance of soft tissue maturation and its influence on facial appearance has been emphasized in orthodontics (59, 60), estimation of apparent age has not been used in the orthodontic literature. The similarity between the groups means that no treatment protocol had a greater aging effect than the other (Table 5).

Long-term soft-tissue cephalometric statuses

There was great similarity among the groups in most soft-tissue cephalometric variables, but, as expected, the four-premolar extraction group had a predominant vertical growth pattern as compared with the other two groups, demonstrated by the significantly longer upper and lower lip lengths, and also by the greater lower anterior face height in comparison with the non-extraction group (Table 6) (61). This shows that the growth pattern has been maintained over the years in the groups, which is also a common occurrence (62). However, only this factor was not enough to influence facial attractiveness as they were similar in the long-term, as demonstrated.

No single soft-tissue cephalometric variable presented significant correlation with facial aesthetics, which demonstrates that facial attractiveness is determined by a set of different facial attributes adequately arranged (42,44).

Clinical implications

The similarity in long-term facial attractiveness among the groups may be consequent to the correct treatment protocol used in each patient. Premolar extractions, when correctly indicated, usually improve facial attractiveness (14,48). The main issue is not about the best treatment, but the conditions under which each one is chosen (17).

Strategies for Class II malocclusion treatment should consider important criteria such as: facial appearance, malocclusion severity, amount of crowding, patient age, available space for third molars, mechanical efficiency, and the challenges offered by each device or treatment protocol.

Conclusions

Based on this appropriately treated Class II sample, it can be concluded that:

• There were no significant differences in long-term facial attractiveness and apparent age, between non-extraction, two- and four-premolar extraction of Class II division 1 malocclusion treatment protocols;

• In the long-term, only upper and lower lip lengths were significantly longer in the four-premolar extraction group than the other two groups and its lower anterior face height was greater than the non-extraction group;

• There was no significant correlation between any individual soft-tissue cephalometric variable and facial attractiveness.

Funding

This work was supported by FAPESP—‘São Paulo Research Foundation’—FAPESP. Authors thank FAPESP for the provided funding.