Prenatal Alcohol Exposure and Dysmorphic Features in Adults at Midlife

ABSTRACT

Background

Dysmorphology examinations have established that fetal alcohol spectrum disorders (FASD) are associated with characteristic physical features from infancy through adolescence. As individuals mature into young adulthood and beyond, the aging process may attenuate characteristic features over time. The purpose of this study was to evaluate FASD‐related dysmorphology at midlife.

Methods

Subjects (N = 192) were recruited from two longitudinal cohorts and from the general community (M _age = 39.5 years, SD_age  = 8.1 years). A majority (n = 132, 68.8%) had an FASD or well‐documented prenatal alcohol exposure (PAE). Standard dysmorphology examinations of 23 physical features were conducted, blind to FASD/PAE status. Participants were stratified into experimental groups according to actual or inferred degree of dysmorphology at the time of their initial examination in childhood or adolescence: high dysmorphology or low dysmorphology. Age‐matched controls with no diagnosed or suspected FASD formed the third experimental group. Hierarchical multiple and logistic regression analyses were conducted, regressing the 23 physical characteristics on age, race/ethnicity, and sex (Block 1), experimental group (Block 2), and the two‐way interactions between experimental group and the demographic variables (Block 3).

Results

Numerous main effects were found for the demographic variables. Controlling for these, main effects for experimental group were found for orbitofrontal circumference, palpebral fissure lengths, philtrum score, vermilion border score, hypoplastic midface, and camptodactyly. Demographics moderated effects of experimental group for philtrum length, philtrum score, and vermilion border score, cardinal features of FASD.

Conclusions

Findings indicate that PAE‐related dysmorphological features persist into adulthood and further into midlife and may remain useful in diagnosis of FASD beyond young adulthood. However, as demographic factors appear to impact some of the cardinal features of FASD, age‐, gender‐, and race‐related norms and reference image sets for the sentinel features of FASD in adults of different ages may be helpful in diagnosing individuals in adulthood and beyond.

Dysmorphological facial features in individuals with prenatal alcohol exposure (PAE), including those with fetal alcohol spectrum disorder (FASD), persist into midlife. Controlling for age, sex, and race/ethnicity, PAE/FASD predicted measures of palpebral fissures, the philtrum, and the upper lip, among other features. With future establishment of norms and reference image sets for sentinel features of FASD in adults of both sexes from various racial/ethnic backgrounds, examining these features may remain useful in the diagnosis of FASD beyond young adulthood.

1. Introduction

Fetal alcohol spectrum disorder (FASD) is an umbrella term encompassing the wide‐ranging effects resulting from prenatal alcohol exposure (PAE). Alcohol is a well‐established teratogen known to impair both physical and neurodevelopmental outcomes (Riley et al. 2011). Notably, the prevalence of FASD may be as high as 1%–5%, nearly twice that of other neurodevelopmental disorders, including autism spectrum disorder (May et al. 2018), yet FASD is less well recognized and often remains undiagnosed or misdiagnosed. Among the many characteristics of FASD, physical features and facial dysmorphology have been extensively studied, particularly in the context of fetal alcohol syndrome (FAS) (del Campo and Jones 2017). These physical manifestations of significant PAE serve as critical diagnostic markers—especially during early childhood and when maternal alcohol exposure has not been confirmed (Hoyme et al. 2016). Although it is well established that PAE exerts lifelong effects, it can be challenging to make an accurate diagnosis of FASD in adulthood. Part of the challenge is attributable to gaps in understanding how dysmorphological features evolve across the lifespan, particularly beyond adolescence, into adulthood, and further into middle age.

With the goal of facilitating early identification of individuals affected by PAE, most research efforts have focused on fully characterizing physical features and facial dysmorphology during infancy and early childhood. These studies have identified a characteristic craniofacial phenotype in individuals with FAS, with “sentinel features” comprising short palpebral fissures, a smooth philtrum, and a thin upper lip (upper vermilion) (Akison et al. 2024), as illustrated in Figure 1. Currently, examination of these features constitutes a critical component of FASD diagnosis (Cook et al. 2016; Hoyme et al. 2016). Importantly, consistent with recognition of the full spectrum of alcohol's effects, not all individuals with PAE present with these physical markers, highlighting variability in phenotypic expression and the influence of factors such as dose, timing, and duration of alcohol exposure, as well as genetic and epigenetic factors (Ramsay 2010). Nevertheless, correlational studies suggest that the severity of PAE‐related dysmorphology predicts the severity of neurobehavioral impairment in childhood (Astley and Clarren 2001; del Campo et al. 2024).

FIGURE 1.

Facial features inspected in a standard dysmorphology examination for FASD.

Recognizing that development continues beyond childhood and that adolescence is a period of rapid growth and physical maturation, during which the dysmorphic features associated with PAE may become less pronounced, Streissguth et al. (1991) conducted a longitudinal study of facial dysmorphology. The study included adolescents aged 12–17 years and adults aged 18–40 years (notably, only 3 subjects were over the age of 30), all of whom had received an FASD diagnosis between 5 and 12 years previously. Among the participants, 70% had been diagnosed with FAS based on characteristic features such as short palpebral fissures, a smooth and/or long philtrum, a thin upper lip, and midface hypoplasia. The remaining 30% had been diagnosed with what was then termed fetal alcohol effects (FAE), indicating a partial FAS phenotype and/or central nervous system dysfunction. The results indicated that, during adolescence, growth of the nose, chin, and mouth areas altered the overall facial appearance in many subjects, thereby attenuating the distinctive facial features characteristic of FAS in adolescents and adults.

Another longitudinal study conducted a 20‐year follow‐up assessment of young adults (mean age 23.4 years) with FASD, who had originally been diagnosed in infancy and childhood as having FAS (59%) or FAE (41%) (Spohr et al. 2007). Overall, results indicated that many individuals across the FASD spectrum exhibited catch‐up growth. For example, only about half of those who had originally presented with microcephaly continued to meet criteria for this feature at follow‐up. Additionally, about three‐quarters of those originally identified with hypoplastic maxilla or midface retained this characteristic into adulthood. Interestingly, a greater proportion of subjects were identified as having a thin upper lip compared to 20 years earlier, suggesting that upper lip thinning may become more pronounced with age.

More recently, a longitudinal study by Jacobson et al. (2021) followed a cohort of individuals born to heavy‐drinking women, along with a control group, across multiple developmental stages: early childhood (approximately 4–6 years), school age (approximately 8–10 years), puberty (approximately 12–14 years), and adolescence (approximately 15–17 years). The findings indicated that a thin vermilion border persisted into adolescence, consistent with the results reported by Spohr et al. (2007). In contrast, the presence of a flat philtrum fluctuated over time, while the prevalence of microcephaly decreased, and palpebral fissures became more elongated as individuals with FASD aged.

Although significant progress has been made in characterizing the physical and facial dysmorphology associated with PAE in childhood, critical knowledge gaps remain. Despite the extensive focus on dysmorphological features associated with PAE in FASD research, the trajectory of these features into middle adulthood and beyond remains unexplored. Of particular relevance, normal aging produces changes in facial morphology. Cross‐sectional studies have shown greater length of the philtrum and thinning of the vermilion border of the upper lip in typically developing adults as they age (Sforza et al. 2010). Given normative age‐related changes, it is possible that the hallmark dysmorphological features of FASD attenuate over time. If the sentinel features of FASD continue to be discernable across the lifespan, taking them into consideration could ultimately help to refine diagnostic criteria for older individuals, who did not have the opportunity to be diagnosed as children. Understanding the long‐term trajectory of dysmorphology in individuals with PAE is an essential first step in determining whether there could be any future role of dysmorphology examination in the diagnosis of middle‐aged adults.

Comparing a control group of individuals without FASD to groups of individuals with FASD or well‐documented PAE with and without childhood dysmorphology, the present study sought to investigate whether dysmorphic features associated with PAE that are identifiable in childhood/adolescence persist into midlife. We hypothesized that significant differences in dysmorphological features would be observed between middle‐aged adults with and without FASD/PAE, but that these differences would be attenuated with age. Therefore, we expected to see both main effects for FASD/PAE and an FASD/PAE by age interaction. Recognizing that sex and race/ethnicity might moderate FASD‐/PAE‐related differences, we examined two‐way interactions between the covariates and experimental group but did not have specific hypotheses related to age, sex, or ethnicity.

2. Methods

2.1. Participants

Between 2018 and 2024, 192 individuals underwent dysmorphology examinations as part of two broader studies examining the long‐term impact of PAE at midlife. These studies were conducted in three locations: Seattle, Washington; Atlanta, Georgia, and Vancouver, British Columbia. Table 1 summarizes participant characteristics.

TABLE 1.

Demographic characteristics of the study sample (N = 192) with univariate p values by study site and experimental group.

	Seattle (n = 95)	Atlanta (n = 42)	Vancouver (n = 55)	p *	NO‐FASD (n = 60)	LO‐DYS (n = 74)	HI‐DYS (n = 58)	p *
Age range, years	29–69	35–43	22–64	—	23–63	22–69	25–57	—
Age, mean years (SD)	40.8 (7.3)	37.8 (1.7)	38.5 (11.3)	0.077	40.0 (8.9)	39.0 (8.8)	39.5 (6.0)	0.771
Height, mean cm (SD)	170.4 (10.0)	166.8 (11.4)	167.5 (10.4)	0.110	172.1 (10.0)	169.6 (9.4)	164.4 (11.0)	< 0.001
Weight, mean kg (SD)	87.3 (22.6)	84.7 (28.2)	77.2 (22.5)	0.047	83.0 (23.0)	91.9 (25.0)	74.4 (20.7)	< 0.001
Sex, % female	46.3%	61.9%	70.4%	0.013	58.3%	55.4%	56.1%	0.941
Race/ethnicity, % (n)
Native American/Indigenous	9.5% (9)	0.0% (0)	48.1% (26)	< 0.001	10.0% (6)	20.3% (15)	24.1% (14)	0.117
Black/African American	5.3% (5)	90.5% (38)	0.0% (0)	< 0.001	18.3% (11)	20.3% (15)	29.3% (17)	0.308
Asian/East Indian	0.0% (0)	0.0% (0)	5.6% (3)	0.021	5.0% (3)	0.0% (0)	0.0% (0)	0.035
White	70.5% (67)	9.5% (4)	35.2% (19)	< 0.001	55.0% (33)	45.9% (34)	41.4% (24)	0.317
More than one of the above	14.7% (14)	0.0% (0)	11.1% (6)	0.034	11.7% (7)	13.5% (10)	5.2% (3)	0.277
Hispanic/Latino	5.3% (5)	0.0% (0)	5.6% (3)	0.300	8.3% (5)	1.4% (1)	3.6% (2)	0.135

Note: Absolute numbers are included for Indigenous and Black categories as these are included in hypothesis testing.

Abbreviations: HI‐DYS, high dysmorphology observed in or inferred from childhood/adolescent examination; LO‐DYS, low dysmorphology observed in or inferred from childhood/adolescent examination; NO‐FASD, no FASD diagnosed or suspected in childhood.

^*Significance values for univariate F‐ or chi‐squared tests, as appropriate.

In Seattle, participants (n = 95) were members of a longitudinal cohort that had been referred as children to the University of Washington for clinical evaluation between 1973 and 1995 and later recruited for participation in research studies. Most children were referred due to known or suspected PAE accompanied by cognitive and/or behavioral problems. The cognitive and physical characteristics associated with FASD were assessed in this cohort when the initial referral was made during childhood. Using the diagnostic terms and criteria of the time, members of this cohort were diagnosed with either FAS or FAE, that is, having effects of PAE but not meeting full criteria for FAS (Streissguth et al. 1985, 1991, 1996). The Seattle control group largely consisted of members of the 1973–1995 longitudinal cohort that had been contemporaneously recruited to be similar to participants with PAE with regard to age, sex, and race/ethnicity. These individuals were either unexposed to alcohol prenatally or had PAE but did not meet criteria for FASD (i.e., FAE or FAS at that time). In some cases, control participants were non‐biological siblings of alcohol affected individuals. Study records and commercially available databases were used to identify longitudinal participants' current contact information, enabling the research team to contact and inform them about the study. Thirteen additional controls were recruited via social media from the greater Seattle area and screened to assure that they were never diagnosed with an FASD and were unlikely to have FASD. Specifically, they were asked: (1) Have you ever been diagnosed with a condition related to prenatal alcohol exposure? [No/Yes] (2) To the best of your knowledge, how much alcohol did your birth mother drink when she was pregnant with you? [None/A Little/A Moderate Amount/A Lot] (3) How certain are you about this? [Not At All/Somewhat/Very Much/Extremely] To be eligible, participants needed to respond No to question 1, None or A Little to question 2, and Very Much or Extremely to question 3.

In Atlanta, participants (n = 42) were members of a longitudinal cohort consisting of individuals born between 1980 and 1986 to women receiving prenatal care at a large, inner‐city hospital (Coles et al. 1985, 2002; Lynch et al. 2017). During that time, alcohol use was assessed in all women seeking prenatal care. Pregnant women who reported drinking at least 1 oz of absolute alcohol per occasion at least once a week while pregnant were recruited for research studies and interviewed to obtain detailed information regarding alcohol consumption. Additional pregnant women who denied drinking any alcohol during pregnancy were recruited as controls. All newborns underwent assessment of medical, behavioral, and dysmorphic outcomes (Coles et al. 1985, 1987). In childhood (Coles et al. 1997), adolescence (Coles et al. 2002), and early adulthood (Lynch et al. 2015), additional information was collected on growth, cognitive ability, and presence of dysmorphic features, allowing the alcohol‐exposed offspring to be classified as either dysmorphic or nondysmorphic. Nondysmorphic means that the individual was alcohol‐exposed but did not have physical features consistent with alcohol exposure at birth or during childhood. That is, they did not exhibit growth restriction and their score on a dysmorphology checklist (Coles et al. 2016) was ≤ 1 standard deviation above the mean of the control group. Notably, members of this cohort were not evaluated formally for diagnoses along the FASD spectrum. For the present study, contact information was obtained from study records and from commercially available databases. Eligible individuals were informed by mail, phone, or email of the opportunity to participate.

In Vancouver, participants (n = 55) were members of a cross‐sectional sample recruited from the community (Bodnar et al. 2023). Recruits self‐identifying as having a diagnosis of FASD were asked to provide details regarding their diagnosis, including the specific subtype of FASD (i.e., FAS, partial FAS, alcohol‐related birth defects, alcohol‐related neurodevelopmental disorder, or a related diagnosis), year of diagnosis, and clinic/doctor who performed the diagnosis. Those who could not provide this information were not invited to participate in the study. Participants with FASD were recruited through social media, advertising at events and conferences related to FASD, and through flyers posted at FASD diagnostic clinics. A matched group of individuals who attested to the best of their knowledge that they were not exposed to alcohol prenatally was also recruited from the community through social media, online advertisements, and posting study flyers in libraries, coffee shops and other community locations.

2.1.1. Stratification

Taking into consideration the heterogeneity of the study groups across the three sites, for the purposes of the analyses presented here, we stratified participants according to three levels. The high dysmorphia group (HI‐DYS) included those diagnosed with FAS in childhood from the Seattle and Vancouver sites and the Alcohol‐Exposed/Dysmorphic group from the Atlanta site. This group would have had the strongest dysmorphological features when first examined/diagnosed. The low childhood dysmorphia group (LO‐DYS) included the FAE group from Seattle, the Alcohol‐Exposed/Nondysmorphic group from Atlanta, and all individuals with FASD diagnoses other than FAS from Vancouver. Members of this group may have had some dysmorphological features, but they would have been below the threshold needed for a full FAS diagnosis. The controls from each site were grouped as having no diagnosed or suspected FASD in childhood/adolescence (NO‐FASD).

2.2. Procedures

The study protocols were approved by the institutional review boards at the University of Washington, Emory University School of Medicine, and University of British Columbia. Participants were accompanied by guardians when appropriate. Informed consent forms were read to individuals who could not complete forms due to reading or cognitive problems. Individuals with legal guardians assented to participate in the study while their guardians provided informed consent. All other participants consented on their own behalf. Participants were reimbursed for their time and travel costs. Broader study procedures have been described in detail elsewhere (Bodnar et al. 2023; Coles et al. 2024).

After completing measures including a demographics questionnaire and other procedures outside the scope of this report, participants' height and weight were measured. Participants were examined using the standard dysmorphology exam employed by studies under the aegis of the Collaborative Initiative on Fetal Alcohol Spectrum Disorders (CIFASD) (Jones et al. 2006; Mattson et al. 2010). This physical examination, which was carried out in person, included both the sentinel features (short palpebral fissures, flattened philtrum, and thin upper vermilion) as well as a number of other features that have been associated with alcohol exposure (see Table 2). Palpebral fissure length (PFL) was measured using a ruler as is most common in clinical examinations. To characterize the philtrum and upper vermilion, examiners employed the lip‐philtrum guides used under the FASD 4‐Digit Code diagnostic approach (Astley 2013). These 5‐point pictorial scales are designed such that higher numbers are more suggestive of prenatal alcohol impacts, with a score of 5 being characteristic of FAS and a score of 3 reflecting the population mean. Due to ethnic differences, there are separate scales for Black and White individuals. As of this writing, the guides are available online at https://depts.washington.edu/fasdpn/htmls/lip‐philtrum‐guides.htm. Examiners were either themselves expert dysmorphologists or trained by expert dysmorphologists. Prior to conducting study‐related examinations, trainees were assessed for inter‐rater reliability with an expert trainer and found to be performing within acceptable limits. Table 2 enumerates the dysmorphology measures examined across the three sites. Figure 1 presents a diagram of facial features inspected in a standard dysmorphology examination for FASD.

TABLE 2.

Descriptive statistics and univariate test statistics for all dysmorphology indices by experimental group. Means for continuous variables are presented in the top panel while proportions for categorical variables are in the bottom panel.

	NO‐FASD		LO‐DYS		HI‐DYS		N	F *	p
	Mean	SD	Mean	SD	Mean	SD
Occipito‐frontal circumference	57.2	2.2	56.6	3.7	54.8	5.0	191	6.69	0.002
Inner canthal distance	3.0	0.3	3.1	0.4	3.0	0.3	192	0.77	0.463
Palpebral fissure length—left	2.9	0.2	2.8	0.2	2.7	0.2	192	9.91	< 0.001
Palpebral fissure length—right	2.8	0.2	2.8	0.2	2.6	0.2	151	12.48	< 0.001
Maxillary arc	28.2	2.2	28.9	3.7	27.8	1.7	150	1.96	0.145
Mandibular arc	30.7	2.4	31.3	3.2	30.1	2.4	150	2.70	0.070
Philtrum length	1.6	0.4	1.7	0.3	1.7	0.3	190	0.48	0.620
Philtrum score	3.1	0.8	3.3	0.7	3.8	0.7	190	13.70	< 0.001
Vermilion border score	2.8	0.8	3.1	0.8	3.5	0.9	190	10.74	< 0.001

	NO‐FASD (%)		LO‐DYS (%)		HI‐DYS (%)		N	x 2 *	p
Hypoplastic midface	17.5		31.5		49.1		187	13.00	0.002
Railroad track configuration of ears	6.7		11.0		8.6		191	0.76	0.685
Strabismus	5.0		5.4		12.1		192	2.81	0.245
Ptosis	5.0		8.1		13.8		192	2.91	0.234
Epicanthal folds	11.7		5.4		5.2		192	2.47	0.290
Anteverted Nares	3.3		13.5		8.9		190	4.22	0.122
Clinodactyly of the fifth Finger	8.3		10.8		15.5		192	1.55	0.461
Camptodactyly	0.0		13.5		13.8		192	9.03	0.011
Difficulty with pronation/supination of elbows	5.0		12.3		15.5		191	3.55	0.170
Contractures in other joints	1.7		4.1		7.3		188	2.17	0.339
Hockey stick upper palmar crease	10.2		9.6		15.5		190	1.28	0.528
Other altered palmar creases	10.3		9.6		17.5		188	2.17	0.339
Thoracotomy scar	0.0		0.0		6.7		125	6.44	0.040
Heart murmur	0.0		1.9		3.4		120	1.22	0.543

Abbreviations: HI‐DYS, high dysmorphology observed in or inferred from childhood/adolescent examination; LO‐DYS, low dysmorphology observed in or inferred from childhood/adolescent examination; NO‐FASD, no FASD diagnosed or suspected in childhood.

^*All between‐groups tests had 2 degrees of freedom.

2.3. Data Analysis

Initial analyses of demographics by experimental group and study site were conducted using F‐tests and chi‐square tests, as appropriate. Means, proportions, and p‐values are presented in Table 1. We also inspected means and proportions for the dysmorphology measures by experimental group, as presented in Table 2.

2.3.1. Hypothesis Testing

We used hierarchical multiple regression (linear or logistic, as appropriate) to test our hypothesis that age would attenuate between‐groups differences in dysmorphological features. To rule out other factors that might affect facial features, sex and race/ethnicity were entered along with age at recent dysmorphology exam in the first block. Race/ethnicity was tested using two dummy coded variables: Black/African American (“Black,” no/yes) and Indigenous/American Indian/Alaska Native (“Indigenous,” no/yes). Experimental group was entered in the second block. Two‐way interactions between the first block variables and experimental group were entered in the third block.

2.3.2. Exploratory Analysis

Beyond hypothesis testing, we were interested in exploring whether a weighted combination of continuous/ordinal dysmorphological features could accurately classify participants in terms of their experimental group membership. We thus performed a discriminant function analysis using only participants in the NO‐FASD and HI‐DYS groups. The LO‐DYS group was excluded from this exploratory analysis due to its heterogeneity. Experimental group was the dependent variable. Continuous variables with significant univariate effects for experimental group as listed in Table 2 were simultaneously entered as predictors. Specifically, we entered OFC, left PFL, philtrum score, and vermilion border score. Right PFL was not entered because it was very highly correlated with left PFL (r = 0.985) and because the Atlanta site did not measure it.

3. Results

Descriptive statistics by group for demographics and all dysmorphology indices are shown in Tables 1 and 2 with univariate p‐values.

3.1. Occipito‐Frontal Circumference (OFC)

As illustrated in Figure 1, OFC is a measure of head circumference. There were significant main effects for Black, β = 0.165, p = 0.026, and sex, β = −0.219, p = 0.002, in block 1, R ² = 0.076, F(4,185) = 3.814, p = 0.005. Black participants and male participants had higher OFCs. There was a significant main effect for experimental group. β = −0.273, p < 0.001, in block 2, ΔR ² = 0.071, F(1,184) = 15.351, p < 0.001. Controlling for race/ethnicity, sex, and age at examination, the effect for experimental group was significant, with higher OFCs measured in the control group compared to the other groups. None of the two‐way interactions entered in block 3 were significant, and the block did not account for a significant proportion of additional variance.

3.2. Inner Canthal Distance (ICD)

ICD is a measure of the distance between the inner corners of two eyes, across the bridge of the nose (see Figure 1). There were significant main effects for Black, β = 0.296, p < 0.001, and Indigenous, β = 0.281, p < 0.001, in block 1, R ² = 0.127, F(4,186) = 6.771, p < 0.001. Black and Indigenous participants had larger ICDs. There were no other significant main effects or significant two‐way interactions.

3.3. Palpebral Fissure Length (PFL)

PFL refers to the distance between the inner and outer corner of the eye (see Figure 1). There were significant main effects for Black, β = 0.294, p < 0.001, Indigenous, β = 0.144, p = 0.045, and sex, β = −0.149, p = 0.033, in block 1, R ² = 0.118, F(4,186) = 6.224, p < 0.001. Black, Indigenous, and male participants had longer left side PFLs. There was a significant main effect for alcohol exposure, β = −0.373, p < 0.001, in block 2, ΔR ² = 0.133, F(1,185) = 32.798, p < 0.001. Controlling for race/ethnicity, sex, and age at examination, longer left side PFLs were noted in the control group compared to the other groups. None of the two‐way interactions entered in block 3 was significant, and the block did not account for a significant proportion of additional variance.

As noted above, right side PFL was not measured at the Atlanta site, decreasing the racial diversity of the data at hand for this variable. There were no significant effects of race/ethnicity, sex, or age at examination for right side PFL in block 1. Controlling for these variables, there was a significant main effect for experimental group, β = −0.419, p < 0.001, in block 2, ΔR ² = 0.168, F(1,144) = 30.812, p < 0.001, similar to that seen with left side PFL. None of the two‐way interactions entered in block 3 was significant, and the block did not account for a significant proportion of additional variance.

3.4. Maxillary Arc

This refers to the length of the U‐shaped arch of the upper jawbone structure that supports the upper teeth (see Figure 1). Maxillary arc was not measured at the Atlanta site, decreasing the racial diversity of the available data for this variable. There were significant main effects for Indigenous, β = 0.320, p < 0.001, and sex, β = −0.223, p = 0.004, in block 1, R ² = 0.159, F(4,144) = 6.823, p < 0.001. Indigenous and male participants had higher values for the maxillary arc. There was no other significant main effect nor were there two‐way interactions.

3.5. Mandibular Arc

This refers to the length of the U‐shaped arch formed by the 16 inferior teeth that are embedded in the lower jaw (see Figure 1). Mandibular arc was not measured at the Atlanta site, decreasing the racial diversity of the data for this variable. There were significant main effects for Indigenous, β = 0.248, p < 0.001, and sex, β = −0.412, p < 0.001, in block 1, R ² = 0.249, F(4,144) = 11.917, p < 0.001. As with maxillary arc, Indigenous and male participants had higher values for the mandibular arc. There were no other significant main effects nor two‐way interactions.

3.6. Philtrum Length

The philtrum is the midline groove that is typically found above the center of the upper lip, extending to the base of the nose, which varies in size and depth (see Figure 1). There were significant main effects for Indigenous, β = 0.150, p = 0.035, Black, β = −0.160, p = 0.025, sex, β = −0.218, p = 0.002, and age at examination, β = 0.147, p = 0.034, in block 1, R ² = 0.147, F(4,184) = 7.920, p < 0.001. Indigenous, male, and older participants had longer philtrums. Black participants had shorter philtrums. Controlling for these variables, there was no significant main effect for experimental group in block 2, but there was a significant interaction between experimental group and Indigenous, β = −0.401, p = 0.005, in block 3, ΔR ² = 0.046, F(4,179) = 2.552, p = 0.041. As shown in Figure 2, the greatest philtrum lengths were observed in Indigenous participants in the control group.

FIGURE 2.

Box‐and‐Whisker plots of lip and philtrum measurements, showing interactions with experimental group membership, with + indicating the group mean.

3.7. Philtrum Score

This validated, 5‐point measure characterizes the dimensions of the philtrum (Astley 2013), with higher scores indicating flatter philtrums. There were significant main effects for sex, β = 0.156, p = 0.031, and age at examination, β = 0.172, p = 0.018, in block 1, R ² = 0.069, F(4,184) = 3.388, p = 0.011. The effect for Black was marginally significant, β = 0.140, p = 0.062. Black, female, and older participants had a smoother philtrum appearance. Controlling for race, sex, and age at examination, there was a significant main effect for experimental group, β = 0.375, p < 0.001, in block 2, ΔR ² = 0.134, F(1,183) = 30.825, p < 0.001, with the smoothest philtrum appearance observed in the HI‐DYS group. Block 3 accounted for a significant proportion of additional variance, ΔR ² = 0.050, F(4,179) = 2.979, p = 0.021. There was a significant interaction between experimental group and sex, β = 0.673, p = 0.004. As shown in Figure 2, the increasing values for philtrum score associated with experimental group membership were observed most clearly in females.

3.8. Vermilion Border Score

This 5‐point measure characterizes the dimensions of the upper lip (Astley 2013), with higher scores reflecting smaller upper lips. There were significant main effects for Indigenous, β = −0.143, p = 0.048, Black, β = −0.162, p = 0.027, and age at examination, β = 0.278, p < 0.001, in block 1, R ² = 0.116, F(4,184) = 6.027, p < 0.001. Indigenous, Black, and younger participants had a larger appearing upper lip. Controlling for race, sex, and age at examination, there was a significant main effect for experimental group, β = 0.368, p < 0.001, in block 2, ΔR ² = 0.129, F(1,183) = 31.314, p < 0.001, with the smallest appearing upper lip in the HI‐DYS group. There were significant interactions between experimental group and Black, β = −0.427, p < 0.001, and experimental group and sex, β = 0.501, p = 0.025, in block 3, ΔR ² = 0.069, F(4,179) = 4.520, p = 0.002. As shown in Figure 2, the increasing values for vermilion border score associated with experimental group membership were observed most clearly in non‐Black and female individuals.

3.9. Hypoplastic Midface

This refers to a condition where the upper jaw, cheekbones, and eye sockets do not grow to the full extent of the rest of the face, creating an appearance of a sunken or flattened midface. Its identification can be somewhat subjective and should be considered relative to the maxillary and mandibular arcs. Block 1 was marginally significant with race/ethnicity, sex, and age in the model, χ ²(4, N = 186) = 9.355, p = 0.053, Nagelkerke R ² = 0.069, correctly classifying 67.2% of cases. Female was associated with a 60.5% decrease in the odds of being classified as having a hypoplastic midface (OR = 0.395, 95% CI [0.21, 0.75]). Block 2 was significant, χ ² (1, N = 186) = 13.015, p < 0.001 with the addition of experimental group to the model, χ ² (5, N = 186) = 22.369, p < 0.001, Nagelkerke R ² = 0.158, correctly classifying 73.1% of cases. For every unit increase in experimental group, there was a 121.3% increase in the odds of being classified as having a hypoplastic midface (OR = 2.213, 95% CI [1.41, 3.47]). Block 3 was not significant with the addition of the two‐way interactions to the model.

3.10. Railroad Track Ears

A parallel appearance of the ridges of the outer ear was observed in 17 individuals, but there were no significant effects of the predictor variables.

3.11. Strabismus

Misalignment of the eyes was observed in 14 individuals, but there were no significant effects of the predictor variables.

3.12. Ptosis

Drooping of the upper eyelid was observed in 17 individuals. Block 1 was significant, χ ² (4, N = 191) = 9.534, p = 0.049, Nagelkerke R ² = 0.108, correctly classifying 91.1% of cases. Black was associated with 455.2% increase in odds of being classified as having ptosis (OR = 5.552, 95% CI [1.73, 17.78]). Block 2 was not significant with the addition of experimental group to the model, and Block 3 was not significant with the addition of the two‐way interactions.

3.13. Epicanthal Folds

Skin folds of the upper eyelid that cover the inner corner (medial canthus) of the eye were observed in 14 individuals. Block 1 was significant, χ ² (4, N = 191) = 13.208, p = 0.010, Nagelkerke R ² = 0.164, correctly classifying 92.7% of cases. However, none of the predictor variables had a significant impact on the odds of classification. With the addition of the experimental group to the model, Block 2 was not significant, but the model remained significant, χ ² (5, N = 191) = 15.611, p = 0.008, Nagelkerke R ² = 0.192. In this block, Indigenous became significant and associated with a 257.3% increase in the odds of being classified as having epicanthal folds (OR = 3.573, 95% CI [1.06, 12.07]). Block 3 was not significant with the addition of the two‐way interactions.

3.14. Anteverted Nares

Upturned nostrils giving the nose a shortened appearance were observed in 17 individuals. There were no significant effects of the predictor variables.

3.15. Fifth Finger Clinodactyly

A sideways curve of the fifth finger was observed in 21 individuals. Block 1 was significant, χ ² (4, N = 191) = 17.373, p = 0.002, Nagelkerke R ² = 0.174, correctly classifying 89.0% of cases. Female was associated with 194.7% increase in odds of being classified as having clinodactyly (OR = 2.947, 95% CI [1.01, 8.57]). Blocks 2 and 3 were not significant with the addition of experimental group and the two‐way interactions to the model.

3.16. Camptodactyly

Flexion contracture at the proximal interphalangeal joint can be an indicator of brain injury. It was observed in 18 individuals. Block 1 was not significant with race/ethnicity, sex, and age in the model. Block 2 was significant, χ ² (1, N = 191) = 7.766, p = 0.005 with the addition of experimental group to the model, χ ² (5, N = 191) = 15.339, p = 0.009, Nagelkerke R ² = 0.166, correctly classifying 90.1% of cases. For every unit increase in experimental group, there was a 175.9% increase in the odds of being classified as having camptodactyly (OR = 2.759, 95% CI [1.28, 5.97]). Additionally, with experimental group in the model, every unit increase in age was associated with 10.0% decrease in odds of camptodactyly (OR = 0.900, 95% CI [0.83, 0.98]). Block 3 was not significant with the addition of the two‐way interactions.

3.17. Difficulty With Pronation/Supination of Elbows

Observed in 21 individuals, there were no significant effects of the predictor variables.

3.18. Contractures in Other Joints

These were observed in eight individuals, but there were no significant effects of the predictor variables.

3.19. Hockey Stick Upper Palmar Crease

An upper palmar crease resembling a hockey stick was observed in 22 individuals. There were no significant effects of the predictor variables.

3.20. Other Altered Palmar Creases

Observed in 23 individuals, there were no significant effects of the predictor variables.

3.21. Thoracotomy Scar

Thoracotomy describes an incision made in the chest wall to access the contents of the thoracic cavity. With only two individuals showing evidence of the procedure, there was insufficient variance to fit a model to the data.

3.22. Heart Murmur

Only two individuals were found to have a current heart murmur. There were no significant effects of the predictor variables.

3.23. Taking Height and Weight Into Consideration

All analyses presented above were re‐run with the addition of height and weight as covariates. Although minor differences in results emerged, there were no differences in main effects for the experimental group. That is, the significant differences between the experimental groups were not simply a function of those with FASD being smaller on average.

3.24. Discriminant Function Analysis

This exploratory analysis examined whether a weighted combination of dysmorphological features (OFC, PFLL, philtrum score, vermilion score) could accurately categorize participants as to their experimental group membership. Although Box's M was significant, M = 59.63, F(10,61,004) = 5.74, p < 0.001, indicating a violation of the assumption of equality of covariance matrices, the group sizes were nearly equal (58 and 57 cases in the NO‐FASD and HI‐DYS groups, respectively). Because Discriminant Function Analysis is robust to this violation when group sizes are balanced (Buyukozturk and Çokluk Bökeoğlu 2008), the analysis proceeded using the within‐groups covariance matrix. Furthermore, a sensitivity analysis using separate‐groups covariance matrices yielded no substantial changes in classification accuracy or predictor importance. We thus present the findings, considering that this analysis was exploratory. The results revealed one significant function, Wilks' λ = 0.682, χ ² (4, N = 115) = 42.503, p < 0.001. This function accounted for 100% of the between‐group variance with a canonical correlation of 0.564. Standardized coefficients indicated that philtrum score (0.587) and vermilion border score (0.384) were the strongest predictors. Left PFL (−0.326) and OFC (−0.288) were also good predictors. Likewise, the structure matrix revealed that higher scores on philtrum score (r = 0.734) and vermilion border score (r = 0.627) were associated with membership in the HI‐DYS group, whereas higher scores on left PFL (r = −0.593) and OFC (r = −0.470) were characteristic of the control group. The model correctly classified 73% of original cases and 70% of cross‐validated cases.

4. Discussion

This study sought to investigate whether the dysmorphic features characteristic of FASD in childhood and adolescence persist into midlife. We hypothesized that significant between‐groups differences in dysmorphological features would be observed, controlling for age, sex, and race/ethnicity, between those with well‐documented PAE and/or childhood diagnosis of FASD and those without diagnosed or suspected FASD. Our hypothesis was supported for several of the features often observed in FASD, including philtrum structure, thinned upper lip (or vermilion), PFLs, OFC, hypoplastic midface, and camptodactyly. Compared to controls, individuals in the FASD/PAE groups had measures indicating significantly smoother philtrums, thinner upper lips, shorter PFLs, smaller OFC, higher likelihood of hypoplastic midface, and higher likelihood of camptodactyly. The study findings suggest that the same cardinal features associated with FASD in childhood/adolescents may be useful in refining the diagnosis of FASD in adulthood in the future.

Importantly, this study did not attempt to diagnose or re‐diagnose FASD in adults. We were fundamentally interested in exploring whether aging would diminish between‐groups differences in physical features that we would expect to see based on well‐documented PAE or FASD diagnosis in childhood/adolescence. Cognizant that gender and ethnic differences might influence the presentation of these features, we included these variables as covariates. Had there been no between‐groups differences, it would have enabled us to say that age‐related attenuation of the sentinel features of FASD in adulthood clearly occurs to the extent that dysmorphology examinations of adults being considered for FASD diagnosis are unlikely to be fruitful. The significant findings presented here suggest that further work to examine the applicability of dysmorphology examinations to adult diagnosis of FASD would be worthwhile.

Main effects for experimental group in the present study were, in some cases, moderated by sex or race/ethnicity, particularly in measures of the lip and philtrum. Increasing values for philtrum score and vermilion border score associated with experimental group membership were observed most clearly in females. Non‐Black individuals most clearly demonstrated increasing values for vermilion border score associated with experimental group membership. Compared to non‐Indigenous individuals, Indigenous individuals with no PAE exhibited the greatest philtrum lengths. In the scientific literature on facial aging, chronological age interacts with race/ethnicity and sex to influence both the pattern and magnitude of morphological changes, although detailed comparative data—especially across diverse racial/ethnic groups—remain relatively limited. Aging is consistently associated with soft‐tissue thinning and volume loss in the perioral area, including the upper lip, as shown in longitudinal MRI studies of adult cohorts, which report significant reductions in upper lip thickness and volume with age in both sexes (e.g., up to ~40% reduction in soft‐tissue thickness in older vs. younger adults), with relatively greater thinning observed in women than men in samples of primarily Caucasian individuals (Ramaut et al. 2019). Studies quantifying upper lip metrics in Asian populations similarly report age‐related decreases in upper‐lip 3D height and volume and flattening of the vermilion border, indicating that lip thinning and elongation are general features of aging across ethnicities, albeit nuanced by population‐specific baseline morphology (Yang et al. 2023).

Race and ethnicity influence overall facial aging trajectories, likely reflecting genetic and structural tissue differences. For example, individuals of African descent generally exhibit slower bone resorption and less pronounced midfacial bony remodeling compared with Caucasians, which may attenuate some age‐related changes in facial support structures that underlie soft‐tissue sagging (Buziashvili et al. 2019). However, quantitative comparative studies specifically measuring upper‐lip thinning across racial and ethnic groups are lacking, and existing anthropometric data primarily describe cross‐sectional differences rather than longitudinal aging effects.

Sexual dimorphism modifies aging phenotypes such that many facial features, including lip vermilion height and overall mouth dimensions, are larger in males across the lifespan, and many age‐related changes (e.g., vermilion narrowing and philtrum elongation) exhibit statistically significant age by sex interactions (Sforza et al. 2010). Moreover, hormonal differences related to menopause and intrinsic differences in dermal collagen and elasticity may contribute to more pronounced or earlier onset of soft‐tissue thinning and volume loss in women versus men, consistent with broader observations of sex differences in aging rates of facial tissues (Velemínská et al. 2022). In sum, while upper lip thinning with aging is a broadly observed phenomenon, its severity and timing vary with sex and, to a lesser understood extent, race/ethnicity, reflecting complex interactions between soft‐tissue biology, underlying skeletal aging, and population‐specific morphological baselines. Our findings and those of others point to the importance of taking sex and race/ethnicity into account when using lip and philtrum guides.

There are several limitations in this study. Low numbers of participants in certain categories could have adversely affected statistical power or resulted in unexpected findings. For example, there were only six Indigenous participants in the control group. It should be noted that our major findings are between‐group findings, and differences in experimental groupings between the sites increased heterogeneity in the experimental groups in the present study. We do not have longitudinal, within‐subjects dysmorphology exam data available to the present study. However, it would seem likely that individuals with observable dysmorphological features in midlife were those that had such features in childhood. Conversely, it would seem unlikely that an individual without dysmorphological features in childhood would develop such features later in life. In the Atlanta sample, participants had been classified as having the presence or absence of dysmorphological features in childhood and the presence or absence of PAE, but they were not formally diagnosed with FASD. By contrast, in the Seattle sample, participants had been identified as having the presence or absence of a FAS or FAE diagnosis in childhood. Presence of dysmorphological features and maternal alcohol use during pregnancy would certainly have factored into these diagnoses. Relatedly, the Vancouver sample was stratified based on diagnosis, which presumably took dysmorphological features and maternal alcohol use during pregnancy into account. We can thus infer that the significant findings in the present study are driven by those who had dysmorphological features in childhood/adolescence, forming part of the basis of an FASD diagnosis for a subset of participants and suggesting that some dysmorphological features persist over the lifespan well into middle age.

We did not ask participants about potentially confounding factors such as history of cosmetic procedures, such as recent Botox use. Other noncongenital factors could also have influenced the prevalence of dysmorphological features. For example, hand injury or a disease process like Dupuytren's contracture could result in a positive finding of camptodactyly in an individual, and this could be confounded with age as the longer one lives, the more opportunity one has to be injured or develop a disease. Our dysmorphology examiners asked about alternative causes of observed dysmorphological features and noted them, but we did not take these notes into consideration in the present study.

These findings may have been influenced by use of a ruler and lip and philtrum guides. Improved techniques could enhance the measurement of subtle features, particularly those involving volumetric changes such as midface hypoplasia, shortened noses, or variations in mandibular size. For example, some studies suggest that advanced imaging methods, such as three‐dimensional craniofacial imaging, may enhance the detection of subtle dysmorphic features in adults (Suttie et al. 2013).

Despite these limitations, our findings indicate that traditional standardized dysmorphology examinations, conducted by trained examiners, do detect dysmorphological differences in middle‐aged adults that are attributable to PAE. Individuals with PAE experience the normal process of aging in their faces. This should be taken into account for diagnosis. Similarly, sex, race, and age variation need to be considered. Our findings begin to suggest how normal processes of aging, as well as the differences associated with sex and race/ethnicity, influence the identification of these features and may support more accurate clinical diagnosis of the effects of PAE.

Further investigation and more sensitive identification of both cardinal and noncardinal features of FASD could potentially benefit diagnosis across the lifespan. Indeed, as demographic factors appear to impact some of the cardinal features of FASD, an expanded set of dysmorphological features could be utilized to aid in the accuracy of diagnosis for individuals in adulthood. Furthermore, if dysmorphology examinations are to be used to refine diagnosis of FASD in adults, alone or in combination, future work will need to include establishment of norms and creation of reference image sets for the sentinel features of FASD in adults of both sexes from various racial/ethnic backgrounds without FASD. Additionally, future rigorous studies of adults with pre‐existing FASD diagnoses will be necessary to specify criteria for deviation from established norms in alcohol‐exposed individuals under consideration for FASD diagnosis.

Ultimately, understanding the long‐term trajectory of dysmorphology in individuals with PAE is essential for refining diagnostic criteria for older individuals, which in turn can help tailor interventions and address the unique needs of this population. By examining the persistence and evolution of a broad range of dysmorphic features, we have sought to help lay the groundwork needed to improve our ability to identify and support individuals affected by PAE, particularly those who were undiagnosed as children and might otherwise remain undiagnosed or misdiagnosed as adults.

Funding

This research was supported by NIH Research Grants U01AA026108 (PI, Claire D. Coles) and U01AA026101 (PI, Joanne Weinberg) funded by the National Institute on Alcohol Abuse and Alcoholism (NIAAA). All or part of this work was done in conjunction with the Collaborative Initiative on Fetal Alcohol Spectrum Disorders (CIFASD), which is funded by grants from the National Institute on Alcohol Abuse and Alcoholism (NIAAA). Additional information about CIFASD can be found at www.cifasd.org. Additional support was provided by the SKK Foundation through a gift to the investigators and the Ann Streissguth, Ph.D. Endowed Professorship of Fetal Alcohol Spectrum Disorders.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Scatterplots for left palpebral fissure length and lipometric measures by age and experimental group membership.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.